阅读说明：本技术 异常诊断系统、异常诊断装置以及数据发送装置 (Abnormality diagnosis system, abnormality diagnosis device, and data transmission device ) 是由 前田才夫 加藤夏洋 于 2021-04-23 设计创作，主要内容包括：本发明提供异常诊断系统、异常诊断装置以及数据发送装置。异常诊断系统应用于具备从燃料箱汲取燃料的燃料泵和供从燃料泵喷出的燃料流动的燃料管的燃料供给系统。在异常诊断系统中,将从燃料供给系统的主开关接通到断开为止的一个行程期间的燃料管内的最低燃料压力和表示记录了该最低燃料压力时的状态的数据作为诊断用数据而存储于存储装置。在异常诊断系统中,执行装置使用存储于存储装置的诊断用数据来判别与燃料管内的燃料压力的下降相关的故障的部位,从而诊断燃料供给系统的异常。(The invention provides an abnormality diagnosis system, an abnormality diagnosis device, and a data transmission device. The abnormality diagnostic system is applied to a fuel supply system including a fuel pump that draws fuel from a fuel tank and a fuel pipe through which the fuel discharged from the fuel pump flows. In the abnormality diagnosis system, the lowest fuel pressure in the fuel pipe during one stroke from the turning on to the turning off of the main switch of the fuel supply system and data indicating a state in which the lowest fuel pressure is recorded are stored as data for diagnosis in a storage device. In the abnormality diagnosis system, the execution device uses the data for diagnosis stored in the storage device to determine the location of a failure related to a drop in fuel pressure in the fuel pipe, thereby diagnosing an abnormality of the fuel supply system.)

1. An abnormality diagnosis system for a fuel supply system, applied to a fuel supply system including a fuel pump for pumping up fuel from a fuel tank and a fuel pipe through which the fuel discharged from the fuel pump flows,

the abnormality diagnosis system is provided with a storage device and an execution device,

the abnormality diagnosis system stores, as data for diagnosis, a lowest fuel pressure in the fuel pipe during one stroke from turning on to turning off of a main switch of the fuel supply system and data indicating a state in which the lowest fuel pressure is recorded in the storage device,

the execution device diagnoses an abnormality of the fuel supply system by determining a location of a failure related to a drop in fuel pressure in the fuel pipe using the data for diagnosis.

2. The abnormality diagnostic system of a fuel supply system according to claim 1,

the diagnostic data includes an elapsed time from the start of the fuel pump as data indicating a state in which the lowest fuel pressure is recorded.

3. The abnormality diagnostic system of a fuel supply system according to claim 2,

the actuator is configured to diagnose an abnormality in the fuel supply system by determining deterioration of an impeller of the fuel pump and an operation failure of a check valve provided in the fuel pipe and opened by a flow of the fuel discharged from the fuel pump, and to close the check valve when the fuel pump is stopped and the supply of the fuel is stopped, as a factor of the abnormality relating to the decrease in the fuel pressure in the fuel pipe.

4. The abnormality diagnostic system of the fuel supply system according to claim 2 or 3, wherein,

the execution device diagnoses an abnormality in the fuel supply system by determining a factor of an abnormality related to a drop in the fuel pressure in the fuel pipe using statistical data obtained by counting the number of times the lowest fuel pressure is recorded for each combination of regions into which various types of data included in the diagnostic data are divided based on the magnitude of the value.

5. The abnormality diagnostic system of a fuel supply system according to claim 4,

the diagnostic data includes a temperature of the fuel as data indicating a state in which the lowest fuel pressure is recorded.

6. The abnormality diagnostic system of a fuel supply system according to claim 5,

the actuator is configured to diagnose an abnormality of the fuel supply system by determining deterioration of an impeller of the fuel pump, a malfunction of a check valve provided in the fuel pipe and opened by a flow of the fuel discharged from the fuel pump, and other abnormality not affected by a temperature of the fuel, as causes of the abnormality related to a decrease in the fuel pressure in the fuel pipe.

7. The abnormality diagnostic system of a fuel supply system according to claim 6,

the storage device stores a learning completion model obtained by machine learning using teacher data obtained by giving the statistical data information indicating the presence or absence of an abnormality related to a drop in fuel pressure in the fuel pipe and the type of a factor of the abnormality as a positive solution label,

the execution device uses the statistical data as an input, and determines a factor of an abnormality related to a drop in fuel pressure in the fuel pipe using the learned model, thereby diagnosing an abnormality of the fuel supply system.

8. The abnormality diagnostic system of a fuel supply system according to claim 7,

the learned model is a decision tree.

9. The abnormality diagnostic system of a fuel supply system according to claim 7 or 8, wherein,

the storage device includes a first storage device mounted on a vehicle and storing the diagnostic data, and a second storage device mounted on a device different from the vehicle and storing the learned model,

the execution device is mounted on a device different from the vehicle together with the second storage device, receives the data for diagnosis stored in the first storage device from the vehicle, creates the statistical data, and determines a factor of an abnormality related to a decrease in fuel pressure in the fuel pipe by the learned model stored in the second storage device using the created statistical data, thereby diagnosing an abnormality of the fuel supply system.

10. A data transmission device constituting the abnormality diagnosis system according to claim 9,

the data transmission device is mounted on a vehicle, and includes the first storage device and a transmitter that transmits the diagnostic data.

11. An abnormality diagnostic device constituting the abnormality diagnostic system according to claim 9,

the abnormality diagnosis device is mounted on a device different from the vehicle, and includes the execution device, the second storage device, and a receiver that receives the data for diagnosis.

Technical Field

The present invention relates to an abnormality diagnosis system, a data transmission device, and an abnormality diagnosis device for a fuel supply system.

Background

The fuel supply system disclosed in japanese patent application laid-open No. 2019-143527 calculates a feedback correction amount set based on a difference between a target fuel pressure and a detected fuel pressure in a state where a pump voltage, a pump current, and a discharge flow rate from a fuel pump are all in a predetermined state. In this fuel supply system, the calculated feedback correction amount is integrated, and when the integrated value of the feedback correction amount is equal to or greater than a threshold value, it is determined that the fuel pump is in a degraded state.

Disclosure of Invention

In the fuel supply system described above, it is determined that the degradation of the fuel pump progresses as the integrated value of the correction amount of the feedback control of the fuel pressure in the specific operation state increases. Therefore, a drop in fuel pressure or the like due to a phenomenon that easily occurs only under specific conditions may be diagnosed as degradation of the fuel pump. In addition, in the case where the state where the detected fuel pressure is lower than the target fuel pressure continues, even if the main cause thereof is not deterioration of the fuel pump, it is diagnosed that the fuel pump has deteriorated.

Means for solving the above problems and the effects thereof will be described below.

An abnormality diagnosis system for a fuel supply system for solving the above problem is applied to a fuel supply system including a fuel pump for pumping up fuel from a fuel tank and a fuel pipe through which the fuel discharged from the fuel pump flows. The abnormality diagnosis system includes a storage device and an execution device, and stores, as data for diagnosis, a lowest fuel pressure in the fuel pipe during one stroke from turning on to turning off of a main switch of the fuel supply system and data indicating a state in which the lowest fuel pressure is recorded. The execution device uses the diagnostic data to determine a location of a failure related to a drop in fuel pressure in the fuel pipe, thereby diagnosing an abnormality of the fuel supply system.

When the fuel pressure in the fuel pipe is lowered while the fuel pump is being operated, there is a possibility that an abnormality occurs in the fuel supply system. According to the above configuration, the presence or absence of an abnormality under a specific condition is not diagnosed, but the abnormality is diagnosed based on the lowest fuel pressure and data indicating a state in which the lowest fuel pressure is recorded. Thus, various abnormalities caused by different factors can be detected.

The data indicating the state in which the lowest fuel pressure is recorded is data indicating the state in which the abnormality is estimated to have occurred. Therefore, if the diagnostic data including the data indicating the state in which the minimum fuel pressure is recorded is used, it is possible to estimate the factor that the fuel pressure in the fuel pipe has recorded the minimum fuel pressure.

Therefore, according to the above configuration for diagnosing an abnormality of the fuel supply system using the data for diagnosis, it is possible to diagnose not only the presence or absence of the abnormality but also a location of a failure related to a drop in the fuel pressure in the fuel pipe.

In one aspect of the abnormality diagnosis system for a fuel supply system, the data for diagnosis includes an elapsed time from activation of the fuel pump as data indicating a state in which a lowest fuel pressure is recorded.

The elapsed time from the activation of the fuel pump until the recording of the lowest fuel pressure is data indicating the relationship between the activation start time of the fuel pump and the time at which the drop in the fuel pressure occurred. According to the above configuration, the portion of the failure related to the drop in the fuel pressure in the fuel pipe can be determined with reference to whether the drop in the fuel pressure occurs immediately after the fuel pump is started or when a certain period of time has elapsed after the fuel pump is started.

In one aspect of the abnormality diagnosis system for a fuel supply system, the actuator determines deterioration of an impeller of the fuel pump and an operation failure of a check valve provided in the fuel pipe and opened by a flow of fuel discharged from the fuel pump, and closes the check valve when the fuel pump is stopped and supply of fuel is stopped, as causes of an abnormality related to a drop in fuel pressure in the fuel pipe, thereby diagnosing the abnormality of the fuel supply system.

The check valve is opened when the fuel pump is operating and the fuel is being discharged from the fuel pump and a flow of the fuel from the fuel pump side toward the fuel injection valve side exists in the fuel pipe. When the check valve is not properly opened due to poor operation of the check valve, a drop in fuel pressure is likely to occur immediately after the fuel pump is started. In contrast, if the impeller of the fuel pump is degraded, the impeller of the fuel pump is deformed while the fuel pump is operated, and the impeller interferes with the housing and is hard to rotate. As a result, the fuel pressure is decreased. Such a drop in fuel pressure due to deterioration of the impeller is likely to occur in a time period later than a time period in which a drop in fuel pressure due to a malfunction of the check valve is likely to occur.

Therefore, when the lowest fuel pressure is recorded after the elapsed time from the start of the fuel pump becomes relatively long, the deterioration of the impeller is more likely to be a factor of the decrease in the fuel pressure than the operation failure of the check valve.

Therefore, if data indicating the relationship between the start time of the fuel pump and the time at which the decrease in the fuel pressure occurs is used as the diagnostic data as in the above-described configuration, it is possible to determine deterioration of the impeller of the fuel pump and an operation failure of the check valve, and diagnose an abnormality in the fuel supply system.

In one aspect of the abnormality diagnosis system for a fuel supply system, the execution device diagnoses the abnormality of the fuel supply system by determining a factor of abnormality related to a drop in fuel pressure in the fuel pipe using statistical data obtained by counting the number of times the lowest fuel pressure is recorded for each combination of regions into which various types of data included in the diagnostic data are divided based on the magnitude of the value.

According to the above configuration, since the diagnosis is performed using the statistical data obtained by performing statistics on the diagnostic data in a plurality of trips, it is possible to perform diagnosis with higher accuracy than the case where diagnosis is performed based on the diagnostic data of only one trip.

In one embodiment of the abnormality diagnosis system of the control device of the fuel supply system, the data for diagnosis includes a temperature of the fuel as data indicating a state in which the lowest fuel pressure is recorded.

The fuel temperature at which the lowest fuel pressure is recorded is data indicating the relationship between the fuel temperature and the drop in fuel pressure. When the fuel pressure is decreased due to an abnormality in the fuel supply system, the degree of influence of the temperature of the fuel on the decrease in the fuel pressure differs depending on the portion where the failure occurs. According to the above configuration, the portion of the failure related to the decrease in the fuel pressure in the fuel pipe can be determined with reference to whether the decrease in the fuel pressure occurs when the temperature of the fuel is low or when the temperature of the fuel is high.

In one aspect of the abnormality diagnosis system for a fuel supply system, the actuator determines deterioration of an impeller of the fuel pump, malfunction of a check valve provided in the fuel pipe and opened by a flow of fuel discharged from the fuel pump, and closes the check valve when the fuel pump is stopped and supply of fuel is stopped, and other abnormality not affected by a temperature of fuel, as causes of abnormality related to a drop in fuel pressure in the fuel pipe, to diagnose the abnormality of the fuel supply system.

When the impeller deteriorates, the impeller deforms with an increase in the temperature of the fuel and the temperature of the impeller, and the impeller and the casing interfere with each other, so that the fuel pressure is likely to decrease. Further, the lower the temperature of the fuel, the more likely the check valve is to cause malfunction. Therefore, by performing the diagnosis using the statistical data obtained by performing the statistics on the diagnostic data including the data in which the temperature of the fuel at the time of the lowest fuel pressure is recorded as in the above-described configuration, it is possible to diagnose whether the drop in the fuel pressure is the deterioration of the impeller affected by the temperature of the fuel, the malfunction of the check valve, or another abnormality not affected by the temperature of the fuel.

The other abnormality that is not affected by the temperature of the fuel includes, for example, a rupture of a fuel pipe.

In one aspect of the abnormality diagnosis system for a fuel supply system, the storage device stores a learning-completed model obtained by machine learning using teacher data, the teacher data being obtained by giving the statistical data information indicating the presence or absence of an abnormality related to a decrease in fuel pressure in the fuel pipe and a type of a factor of the abnormality as a positive solution label, and the execution device diagnoses the abnormality of the fuel supply system by using the learning-completed model and by using the statistical data as an input to determine the factor of the abnormality related to the decrease in fuel pressure in the fuel pipe.

It is possible to create a model for diagnosing the presence or absence of an abnormality from statistical data and outputting a type of a factor of the abnormality when the abnormality has occurred, using machine learning.

If machine learning is used, it is possible to extract features that are difficult for a person to perceive and perform abnormality diagnosis. In addition, in the above configuration, if the learned model is a model in which statistical data obtained by performing statistics on the diagnostic data is used as an input, the amount of input data can be reduced as compared with a case where a learned model in which a plurality of diagnostic data themselves are used as inputs is constructed.

In one embodiment of the abnormality diagnosis system for a fuel supply system, the learned model is a decision tree.

Compared with models such as a neural network and the like, the decision tree is easy for people to understand the basis of diagnosis of the learned models. With the above configuration, it is possible to construct an abnormality diagnosis system in which the reason for deriving the diagnosis result can be easily explained.

In one aspect of the abnormality diagnosis system for a fuel supply system, the storage device includes a first storage device mounted on a vehicle and storing the diagnostic data, and a second storage device mounted on a device different from the vehicle and storing the learned model, and the execution device is mounted on a device different from the vehicle together with the second storage device, receives the diagnostic data stored in the first storage device from the vehicle, creates the statistical data, and determines a factor of an abnormality related to a decrease in fuel pressure in the fuel pipe from the learned model stored in the second storage device using the created statistical data, thereby diagnosing the abnormality of the fuel supply system.

With this configuration, the generation of statistical data, the diagnosis of abnormality by the learned model, and the like are performed in a facility different from the vehicle. Therefore, an increase in the capacity of the storage device on the vehicle side and an increase in the calculation load on the vehicle side can be suppressed.

As a device constituting the abnormality diagnosis system having such a configuration, a data transmission device mounted in a vehicle and including a first storage device and a transmitter for transmitting data for diagnosis, and an abnormality diagnosis device mounted in a device different from the vehicle and including an execution device, a second storage device, and a receiver for receiving data for diagnosis can be considered.

Drawings

The features, advantages and technical and industrial significance of exemplary embodiments of the present invention will be described below with reference to the accompanying drawings, in which like reference numerals represent like elements, and in which:

fig. 1 is a schematic diagram showing a relationship between an abnormality diagnostic system of an embodiment and a fuel supply system as a diagnostic target of the abnormality diagnostic system.

Fig. 2 is a schematic diagram showing a change in the state of the impeller in the fuel pump.

Fig. 3 is a table illustrating statistical data for counting the number of times the lowest fuel pressure is recorded, based on the combination of the lowest fuel pressure and the fuel temperature.

Fig. 4 is a table illustrating statistical data for counting the number of times the lowest fuel pressure is recorded based on the combination of the lowest fuel pressure and the elapsed time from the activation of the fuel pump.

Fig. 5 is a decision tree used in the diagnostic process.

Fig. 6 is a flowchart showing a flow of a series of processing in a routine related to acquisition of diagnostic data.

Fig. 7 is a flowchart showing a flow of a series of processes in the routine involved in the update of the statistical data.

Fig. 8 is a flowchart showing a flow of a series of processes in the routine involved in the diagnosis process.

Fig. 9 shows a neural network used for diagnosis processing in the abnormality diagnosis system according to the modified example.

Fig. 10 is a decision tree used for diagnostic processing in the abnormality diagnostic system according to another modification.

Detailed Description

An embodiment of an abnormality diagnosis system for a fuel supply system will be described below with reference to fig. 1 to 8.

Fig. 1 shows the structure of an abnormality diagnostic system and a fuel supply system to which the abnormality diagnostic system is applied according to the present embodiment. The abnormality diagnostic system 600 of the present embodiment is applied to a fuel supply system 550 of an in-vehicle engine mounted on a vehicle 500.

As shown in fig. 1, the fuel supply system 550 to which the abnormality diagnostic system 600 is applied is provided with 2 fuel pumps, i.e., a fuel pump 52 provided in the fuel tank 51 and a high-pressure fuel pump 60 provided outside the fuel tank 51. The fuel pump 52 is an electric pump that rotates an impeller by a brushless motor. Further, the fuel supply system 550 is provided with the in-cylinder fuel injection valves 44 and the port fuel injection valves 30. The in-cylinder fuel injection valve 44 is provided in each cylinder of the engine, and directly injects fuel into the cylinder. The in-cylinder fuel injection valves 44 are connected to a high-pressure side delivery pipe 71 as an accumulator of fuel. Further, port fuel injection valve 30 injects fuel into an intake port connected to each cylinder of the engine. Port fuel injection valve 30 is connected to low-pressure side delivery pipe 31. The engine on which the fuel supply system 550 is mounted is an engine having 4 cylinders arranged in series, and 4 in-cylinder fuel injection valves 44 are connected to the high-pressure side delivery pipe 71. Further, 4 port fuel injection valves 30 are also connected to the low-pressure side delivery pipe 31.

Fuel pipe 57, which is a fuel passage for delivering fuel from fuel pump 52 to high-pressure fuel pump 60 and low-pressure side delivery pipe 31, and high-pressure fuel pipe 72, which is a fuel passage for delivering fuel from high-pressure fuel pump 60 to high-pressure side delivery pipe 71, are provided in fuel supply system 550. The fuel pipe 57 branches off at an intermediate point, and one of the branches is connected to the high-pressure fuel pump 60, and the other branch is connected to the low-pressure side delivery pipe 31.

The low-pressure side delivery pipe 31 is provided with a fuel pressure sensor 131 that detects the pressure of the fuel in the fuel pipe 57 and the low-pressure side delivery pipe 31. Further, the high-pressure side delivery pipe 71 is provided with a fuel pressure sensor 132 that detects the pressure of the fuel accumulated therein, i.e., the high-pressure side fuel pressure. The fuel pressure sensors 131 and 132 indicate the fuel pressure by a gauge pressure based on the atmospheric pressure.

The fuel pump 52 sucks fuel in the fuel tank 51 through the upstream filter 53 and sends the fuel to the fuel pipe 57 in response to the power supply. A relief valve 56 is provided in a portion of the fuel pipe 57 that is positioned inside the fuel tank 51, and the relief valve 56 opens when the pressure of the fuel delivered from the fuel pump 52 to the fuel pipe 57, that is, a feed pressure Pf that is the fuel pressure inside the fuel pipe 57 exceeds a predetermined valve opening pressure, and releases the fuel from the fuel pipe 57 to the fuel tank 51.

Further, a check valve 59 is provided in a portion of the fuel pipe 57 on the upstream side of the portion where the relief valve 56 is provided, the check valve 59 is disposed so that the fuel pump side is downward, and the valve body is seated on a valve seat located downward by its own weight and opens due to the flow of the fuel discharged from the fuel pump 52. When fuel pump 52 is stopped and the supply of fuel is stopped, check valve 59 is closed.

The fuel pipe 57 is connected to a high-pressure fuel pump 60 via a downstream-side filter 58 that filters impurities in the fuel flowing through the fuel pipe 57 and a pulsation damper 61 that reduces pulsation of the fuel pressure in the fuel pipe 57.

The high-pressure fuel pump 60 includes a plunger 62, a fuel chamber 63, an electromagnetic spill valve 64, a check valve 65, and a spill valve 66. The plunger 62 is driven to reciprocate by a pump cam 67 provided on the camshaft 42 of the engine, and the volume of the fuel chamber 63 is changed by this reciprocating drive. The fuel chamber 63 is connected to the fuel pipe 57 via an electromagnetic spill valve 64.

The electromagnetic spill valve 64 closes upon energization to block the flow of the fuel between the fuel chamber 63 and the fuel pipe 57, and opens upon stoppage of energization to permit the flow of the fuel between the fuel chamber 63 and the fuel pipe 57. The check valve 65 permits the fuel to be discharged from the fuel chamber 63 to the high-pressure side delivery pipe 71, while prohibiting the fuel from flowing backward from the high-pressure side delivery pipe 71 to the fuel chamber 63. The relief valve 66 is provided in a passage bypassing the check valve 65, and opens when the pressure on the high-pressure side delivery pipe 71 side becomes excessively high to allow the fuel to flow backward toward the fuel chamber 63.

The fuel pressurizing operation of the high-pressure fuel pump 60 configured as described above will be described. In the high-pressure fuel pump 60, the volume of the fuel chamber 63 changes in accordance with the reciprocating motion of the plunger 62. In the following description, the operation of the plunger 62 in the direction in which the volume of the fuel chamber 63 increases is referred to as the downward movement of the plunger 62, whereas the operation of the plunger 62 in the direction in which the volume of the fuel chamber 63 decreases is referred to as the upward movement of the plunger 62.

In the high-pressure fuel pump 60, when the plunger 62 starts to descend with the electromagnetic spill valve 64 open, the fuel flows into the fuel chamber 63 from the fuel pipe 57 in accordance with the expansion of the volume of the fuel chamber 63. When the electromagnetic spill valve 64 is also kept open after the plunger 62 is shifted from the lower position to the upper position, the fuel that has flowed into the fuel chamber 63 during the lowering of the plunger 62 is returned to the fuel pipe 57. When the electromagnetic spill valve 64 is closed during the ascent of the plunger 62 and thereafter the electromagnetic spill valve 64 is maintained closed until the plunger 62 transitions from the ascent to the descent, the fuel in the fuel chamber 63 is pressurized by the reduction in volume of the fuel chamber 63 accompanying the ascent of the plunger 62. When the fuel pressure in the fuel chamber 63 exceeds the fuel pressure in the high-pressure fuel pipe 72, the check valve 65 opens, and the pressurized fuel in the fuel chamber 63 is sent to the high-pressure fuel pipe 72. Thus, the high-pressure fuel pump 60 pressurizes the fuel in the fuel pipe 57 and sends the fuel to the high-pressure fuel pipe 72 for each reciprocation of the plunger 62. By changing the closing time of electromagnetic spill valve 64 during the raising of plunger 62, the amount of fuel delivered to high-pressure fuel pipe 72 by high-pressure fuel pump 60 is increased or decreased for each pressurizing operation.

The engine provided with such a fuel supply system 550 is controlled by the control device 100. The control device 100 is a control device for the engine, and is also responsible for controlling the fuel supply system 550 of the engine. That is, the control device 100 is also a control device of the fuel supply system 550.

The control device 100 includes an execution device 101 that executes various arithmetic processes, and a storage device 102 that stores control programs and data. The control device 100 includes a transmitter 103 that transmits data via the communication network 400 and a receiver 104 that receives data via the communication network 400.

The control device 100 reads and executes a program stored in the storage device 102 by the execution device 101, and controls the engine including the control of the fuel supply system 550.

Detection signals of various sensors for detecting the operating state of the engine are input to the control device 100. As shown in fig. 1, a detection signal of the amount of operation of the accelerator by the driver is input to the control device 100 from an accelerator position sensor 142, and a detection signal of the vehicle speed, which is the traveling speed of the vehicle, is input from a vehicle speed sensor 141.

In addition, detection signals of various sensors are input to the control device 100. For example, as shown in fig. 1, an air flow meter 133, a crank position sensor 134, a cam position sensor 135, and a cooling water temperature sensor 136 are connected to the control device 100 in addition to the fuel pressure sensors 131 and 132.

The air flow meter 133 detects the temperature of air taken into the cylinder through the intake passage of the engine and the mass of the air taken in, that is, the intake air amount. The crank position sensor 134 outputs a crank angle signal corresponding to a change in the rotational phase of the crankshaft, which is the output shaft of the engine. Control device 100 calculates the engine speed, which is the rotational speed of the crankshaft per unit time, based on the crank angle signal input from crank position sensor 134.

The cam position sensor 135 outputs a cam angle signal corresponding to a change in the rotational phase of the camshaft 42. The cooling water temperature sensor 136 detects a cooling water temperature that is the temperature of the cooling water of the engine.

Also connected to the control device 100 are a fuel temperature sensor 137 that detects the fuel temperature Tf, which is the temperature of the fuel in the fuel tank 51, a fuel level sensor 138 that detects the level of the liquid level of the fuel in the fuel tank 51 and outputs a detection signal indicating the remaining amount of the fuel, and an outside air temperature sensor 139 that detects the outside air temperature.

The control device 100 is also connected to a main switch 140 and a display unit 150 of the vehicle. When an abnormality occurs in the vehicle 500, the display unit 150 displays an icon or a sentence for reporting the occurrence of the abnormality to the occupant.

Further, the control device 100 is connected to a fuel pump control device 200 that controls the pump rotation speed Np, which is the rotation speed per unit time of the impeller of the fuel pump 52. The fuel pump control device 200 adjusts the electric power supplied to the fuel pump 52 by pulse width modulation based on a command from the control device 100, thereby increasing or decreasing the pump rotation speed Np. The fuel pump control device 200 transmits information on the pump current Ip and the pump rotation speed Np, which are currents supplied to the fuel pump 52, to the control device 100.

As a part of the engine control, the control device 100 performs fuel injection amount control, fuel pressure variable control, and feed pressure control.

In controlling the fuel injection amount, the control device 100 first calculates the required injection amounts, which are the required values of the fuel injection amounts of the in-cylinder fuel injection valve 44 and the port fuel injection valve 30, respectively, in accordance with the engine operating state such as the engine speed and the engine load factor. Next, the control device 100 calculates the valve opening times of the in-cylinder fuel injection valve 44 and the port fuel injection valve 30, which are required for fuel injection of the requested injection amount. Then, the control device 100 operates the in-cylinder fuel injection valve 44 and the port fuel injection valve 30 of each cylinder to inject fuel during a period corresponding to the calculated valve opening time. Further, as part of the fuel injection control, the control device 100 also performs the following fuel cut control: in deceleration in which the accelerator operation amount is "0", fuel injection is stopped to stop the supply of fuel to the combustion chamber of the engine, thereby reducing the fuel consumption rate. When the injection of the fuel is stopped, control device 100 stops the operation of fuel pump 52.

In the variable fuel pressure control, the control device 100 calculates a target value of the high-pressure side fuel pressure based on the load factor of the engine and the like. The target value of the high-pressure-side fuel pressure is basically set to a low pressure when the load factor of the engine is low, and set to a high pressure when the load factor of the engine is high. Then, the control device 100 adjusts the fuel delivery amount of the high-pressure fuel pump 60 in order to reduce the deviation between the detection value of the high-pressure side fuel pressure by the fuel pressure sensor 132 and the target value of the high-pressure side fuel pressure. Specifically, when the detected value of the high-pressure-side fuel pressure is lower than the target value, the closing time of the electromagnetic spill valve 64 during the period in which the plunger 62 is rising is advanced, and the fuel delivery amount of the high-pressure fuel pump 60 is increased. When the detected value of the high-pressure side fuel pressure is higher than the target value, the closing time of the electromagnetic spill valve 64 during the rise of the plunger 62 is delayed, and the fuel delivery amount of the high-pressure fuel pump 60 is reduced.

Next, details of the pressure adjustment processing performed as one loop of the feed pressure control will be described. The pressure adjustment processing is performed for the following purpose. When the fuel fed from the fuel pump 52 and flowing through the fuel pipe 57 is heated to a high temperature by the heat of the engine, vapor is generated in the fuel pipe 57, and the fuel may be supplied to the high-pressure side delivery pipe 71 and the low-pressure side delivery pipe 31 in a stopped state. Since the higher the pressure of the fuel, the higher the vaporization temperature of the fuel, in order to prevent the generation of vapor in the fuel pipe 57, the fuel delivery amount of the fuel pump 52 to the fuel pipe 57 may be increased to increase the feed pressure Pf. However, if the fuel delivery amount is increased, the amount of power consumption of the fuel pump 52 increases accordingly. Then, in the pressure adjustment process, in order to maintain the feed pressure Pf at a low pressure within a range in which the generation of vapor can be prevented, the fuel ejection amount of the fuel pump 52 is adjusted, whereby the power consumption is suppressed and the generation of vapor is prevented.

Specifically, the actuator 101 of the control device 100 calculates a required feed pressure Pf, which is a target value of the feed pressure Pf, based on the fuel temperature Tf detected by the fuel temperature sensor 137. In this control device 100, the required feed pressure Pf is switched in accordance with the fuel temperature Tf. In the control device 100, the required feed pressure Pf is made higher as the fuel temperature Tf is higher, in such a manner that the required feed pressure Pf is not lower than the saturated vapor pressure even when the fuel having the highest saturated vapor pressure among the fuels supposed to be used is used. Then, the actuator 101 calculates the required pump rotation speed Np, which is the target value of the pump rotation speed Np, based on the fuel injection amount Qf and the required feed pressure Pf.

The fuel injection amount Qf can be grasped on the basis of the sum of the required fuel injection amount from the in-cylinder fuel injection valve 44 and the required fuel injection amount from the port fuel injection valve 30, which is the required injection amount calculated by the fuel injection amount control.

In the control device 100, the execution device 101 calculates, as the required pump rotation speed Np, the pump rotation speed Np required to achieve the required feed pressure Pf in consideration of the fuel consumption amount of the execution of the fuel injection control. Specifically, the execution device 101 calculates the required pump rotation speed Np by referring to the calculation map stored in the storage device 102. The operation map is created, for example, so as to be able to calculate the required pump rotation speed Np based on the results of an experiment using gasoline as fuel. In this calculation map, the higher the required feed pressure Pf is, the higher the fuel injection amount Qf is, the higher the required pump rotation speed Np is outputted.

The actuator 101 of the control device 100 calculates the correction amount Δ N of the required pump rotation speed Np based on the required feed pressure Pf and the feed pressure Pf detected by the fuel pressure sensor 131. Specifically, when the feed pressure Pf is smaller than the required feed pressure Pf, the actuator 101 increases the correction amount Δ N by a predetermined amount. On the other hand, the actuator 101 decreases the correction amount Δ N by a predetermined amount when the feed pressure Pf is greater than the required feed pressure Pf. Then, the execution device 101 adds the calculated correction amount Δ N to the required pump rotation speed Np to correct the required pump rotation speed Np. Thus, the required pump rotation speed Np corrected by the correction amount Δ N is input to the fuel pump control device 200. Then, fuel pump control device 200 controls the supply of electric power to fuel pump 52 so as to achieve the input required pump rotation speed Np.

If the required pump rotation speed Np is increased, the amount of fuel discharged from the fuel pump 52 per unit time increases, and therefore the feed pressure Pf becomes high. On the other hand, if the required pump rotation speed Np is decreased, the amount of fuel discharged from the fuel pump 52 per unit time decreases, so the feed pressure Pf becomes low.

In this way, in the fuel supply system 550, the feed pressure Pf is feedback-controlled. The actuator 101 of the control device 100 controls the supply of electric power to the fuel pump 52 so as to achieve the required feed pressure Pf by such pressure adjustment processing.

However, if an abnormality occurs in the fuel supply system 550, the required feed pressure Pf cannot be achieved. Therefore, in the abnormality diagnosis system 600, the server device 300 connected to the control device 100 via the communication network 400 collects information of the fuel supply system 550, and diagnoses an abnormality of the fuel supply system 550.

As shown in fig. 1, the server device 300 includes an execution device 301 and a storage device 302 in which a program and data for control are stored. The server device 300 includes a transmitter 303 that transmits data to the receiver 104 of the control device 100 via the communication network 400, and a receiver 304 that receives data transmitted from the transmitter 103 of the control device 100 via the communication network 400.

Note that, if an abnormality occurs in the fuel supply system 550, the feed pressure Pf decreases, and the feed pressure Pf may become lower than the required feed pressure Pf. In the abnormality diagnostic system 600, the site of the failure that causes such a drop in the feed pressure Pf is discriminated from the 3 classifications determined in advance, thereby diagnosing an abnormality of the fuel supply system 550.

Specifically, in the abnormality diagnosis system 600, the execution device 301 of the server device 300 diagnoses whether or not an abnormality has occurred in the fuel supply system 550. When it is diagnosed that an abnormality has occurred, the execution device 301 determines which of the deterioration of the impeller of the fuel pump 52, the malfunction of the check valve 59, and other abnormalities not affected by the fuel temperature Tf is the cause of the abnormality, and diagnoses the occurrence of the abnormality.

As shown in fig. 2, the fuel pump 52 pumps up fuel by driving an impeller 52c housed in a casing 52b by a brushless motor 52 a. The impeller 52c is made of resin, and if it is deteriorated by the fuel impregnation, it is deformed with the temperature increase. As shown by the two-dot chain line in fig. 2, when the impeller 52c deforms and warps, the impeller 52c interferes with the casing 52b and is hard to rotate. As a result, a drop in the feed pressure Pf occurs. That is, a drop in the feed pressure Pf caused by such deterioration of the impeller 52c is likely to occur when the fuel temperature Tf is high.

As shown in fig. 1, the check valve 59 is disposed downstream of the fuel pump 52. When the check valve 59 is not properly opened due to the malfunction of the check valve 59, the feed pressure Pf is also decreased. Such malfunction of the check valve 59 is likely to occur when the fuel temperature Tf is low or immediately after the start of the operation of the fuel pump 52.

In this way, the decrease in the feed pressure Pf caused by the deterioration of the impeller 52c and the decrease in the feed pressure Pf caused by the malfunction of the check valve 59 are affected by the fuel temperature Tf. On the other hand, there is also a tendency that the frequency of occurrence does not change depending on whether the fuel temperature Tf is high or low, and there is an abnormality that occurs without being affected by the fuel temperature Tf. In the abnormality diagnosis system 600, the occurrence frequency of the drop in the fuel pressure in the fuel supply system 550 is counted, and the execution device 301 classifies such an abnormality that is not affected by the fuel temperature Tf as another abnormality. Note that other abnormalities that are not affected by the fuel temperature Tf include, for example, rupture of the fuel pipe 57.

Specifically, in the abnormality diagnosis system 600, the control device 100 mounted on the vehicle 500 monitors the feed pressure Pf during one trip from the on to the off of the main switch 140, and stores the lowest fuel pressure during the one trip and data indicating the state in which the lowest fuel pressure is recorded in the storage device 102 as data for diagnosis.

As data indicating the state in which the lowest fuel pressure in the stroke is recorded, the control device 100 stores, in the storage device 102, the fuel temperature Tf at the time when the lowest fuel pressure is recorded and the elapsed time from the start of the fuel pump 52 to the time when the lowest fuel pressure is recorded in addition to the value of the lowest fuel pressure. In the abnormality diagnosis system 600, when the main switch 140 is turned off and one trip is completed, the control device 100 transmits the data for diagnosis to the server device 300 via the transmitter 103.

The server device 300 receives the diagnostic data transmitted from the control device 100 via the receiver 304. Then, the server 300 stores the received data for diagnosis in the storage device 302, performs statistics, and creates statistical data used for diagnosis processing. That is, in the server apparatus 300, the execution apparatus 301 updates the statistical data stored in the storage apparatus 302 each time the receiver 304 receives the data for diagnosis. Such statistical data is created for each vehicle 500, that is, for each fuel supply system 550. Then, the fuel supply system 550 of each vehicle 500 is subjected to the diagnostic processing using the created statistical data created for each vehicle 500.

In the abnormality diagnostic system 600, first statistical data in which the number of times the lowest fuel pressure is recorded is counted from the combination of the lowest fuel pressure and the fuel temperature Tf as shown in fig. 3 and second statistical data in which the number of times the lowest fuel pressure is recorded is counted from the combination of the lowest fuel pressure and the elapsed time from the start of the fuel pump 52 as shown in fig. 4 are created.

As shown in fig. 3, in the abnormality diagnostic system 600, the first statistical data is divided into 4 regions of "0 kPa to 99 kPa", "100 kPa to 199 kPa", "200 kPa to 299 kPa", and "300 kPa", according to the magnitude of the value. In addition, as shown in fig. 3, in the abnormality diagnostic system 600, the fuel temperature Tf is divided into 4 regions of "a first temperature region temp 1", "a second temperature region temp 2", "a third temperature region temp 3", and "a fourth temperature region temp 4" in the first statistical data according to the magnitude of the value thereof. It should be noted that the "first temperature region temp 1" is a temperature region corresponding to the case where the fuel temperature Tf at the time of recording the lowest fuel pressure is lower than 30 ℃, for example. The "second temperature region temp 2" is a temperature region corresponding to the case where the fuel temperature Tf at the time of recording the lowest fuel pressure is, for example, 30 ℃ or more and less than 40 ℃. The "third temperature region temp 3" is a temperature region corresponding to the case where the fuel temperature Tf at the time of recording the lowest fuel pressure is, for example, 40 ℃ or more and less than 50 ℃. The "fourth temperature region temp 4" is a temperature region corresponding to the case where the fuel temperature Tf at the time of recording the lowest fuel pressure is, for example, 50 ℃.

As shown in fig. 3, the first statistical data corresponding to the combination of the lowest fuel pressure and the fuel temperature Tf is divided into 16 regions obtained by multiplying the 4 regions corresponding to the magnitude of the value of the lowest fuel pressure and the 4 regions corresponding to the magnitude of the value of the fuel temperature Tf, and the number of times the lowest fuel pressure is recorded is counted.

For example, in the abnormality diagnosis system 600, when the lowest fuel pressure in the diagnostic data newly received by the server device 300 is 300kPa or more and the fuel temperature Tf at which the lowest fuel pressure is recorded is 45 ℃, the number of times in the region of "temp 3 d" is increased by 1 and the statistical data is updated.

As shown in fig. 4, in the abnormality diagnostic system 600, the elapsed time from the start of the fuel pump 52 in each trip is divided into 4 regions of "first time region time 1", "second time region time 2", "third time region time 3", and "fourth time region time 4" in the second statistical data according to the magnitude of the value. The "first time zone time 1" is a time zone corresponding to the elapsed time when the lowest fuel pressure is recorded, for example, less than 10 seconds. The region after the "second time region time 2" is a time region in which the range of the corresponding elapsed time is large, for example, several tens of seconds. That is, the "second time zone time 2" is a time zone in which the elapsed time when the lowest fuel pressure is recorded is 10 seconds or more and less than several tens of seconds, and the "third time zone time 3" is a temperature zone corresponding to a range from the "second time zone time 2" to several tens of seconds later. The "fourth time zone time 4" is a time zone corresponding to a case where the elapsed time when the lowest fuel pressure is recorded is longer than the elapsed time corresponding to the "third time zone time 3".

As shown in fig. 4, the second statistical data corresponding to the combination of the lowest fuel pressure and the elapsed time is divided into 16 regions obtained by multiplying 4 regions corresponding to the magnitude of the value of the lowest fuel pressure and 4 regions corresponding to the magnitude of the value of the elapsed time, and the number of times the lowest fuel pressure is recorded is counted.

For example, in the abnormality diagnosis system 600, when the lowest fuel pressure in the newly received data for diagnosis by the server device 300 is 250kPa and the elapsed time when the lowest fuel pressure is recorded is 15 seconds, the number of times in the region of "time 2 c" is increased by 1 to update the statistical data.

The server device 300 performs a diagnosis process on the fuel supply system 550 using the statistical data every time the diagnostic data on the fuel supply system 550 to be diagnosed is received a predetermined number of times. The predetermined number of times is, for example, several tens to several hundreds of times. That is, in the abnormality diagnosis system 600, the abnormality diagnosis of the fuel supply system 550 in the vehicle 500 is performed every time the number of strokes in the vehicle 500 reaches several tens to several hundreds of strokes.

Specifically, when the server device 300 receives the predetermined number of times of the diagnosis data that has not been reflected in the diagnosis process, the first statistical data and the second statistical data are formed as the input data. Then, the server device 300 inputs the formed input data to the learned model generated by machine learning, and executes the diagnosis process of the fuel supply system 550 using the learned model.

As shown in fig. 5, the learned model in the abnormality diagnosis system 600 is a decision tree and is stored in the storage device 302 in advance. The learning-completed model is obtained by machine learning using teacher data that is converted into input data having a generation ratio obtained by dividing the number of occurrences of each region stored in statistical data by the total number of diagnostic data counted up to that time, and by applying a forward label to the input data. The positive label is information indicating the presence or absence of an abnormality related to the decrease in the feed pressure Pf and the type of the cause of the abnormality. Specifically, the teacher data is given with any of "normal" indicating no abnormality, "impeller" indicating deterioration of the impeller 52c, "check valve" indicating malfunction of the check valve 59, and "other" indicating another abnormality not affected by the fuel temperature Tf.

That is, the input data is composed of 32 numerical values, the 32 numerical values being composed of 16 numerical values respectively indicating a ratio of the number of times the lowest fuel pressure is recorded in the 16 respective regions in the first statistical data and 16 numerical values respectively indicating a ratio of the number of times the lowest fuel pressure is recorded in the 16 respective regions in the second statistical data. The teacher data is composed of 37 values obtained by adding 1 value indicating any one of the above-mentioned 4 tags to these 32 values.

The generation of the learning completion model, i.e., learning, is performed by inputting a set of teacher data, i.e., a data set, to the computer. The computer generates a decision tree so that branches with a larger amount of information obtained are arranged at the top of the tree by repeating a search using a general algorithm used for learning a decision tree and using an input data set by a greedy algorithm. The storage device 302 of the server device 300 stores, as a learned model, a decision tree generated by learning in advance.

As shown in fig. 5, the decision tree stored in the storage device 302 of the abnormality diagnostic system 600 is composed of nodes that branch off a value included in input data according to whether or not the value exceeds a threshold value, and leaves that represent diagnostic results.

In this decision tree, first, at the node N100, it is determined by the execution means 301 whether or not the value of "temp 2 d" in the input data is larger than the threshold value X1. When the value of "temp 2 d" in the input data is greater than the threshold value X1, the node N111 is entered, and the execution device 301 determines whether or not the value of "temp 1 c" in the input data is greater than the threshold value X2. The threshold value X2 is a value smaller than the threshold value X1.

If it is determined at the node N111 that the value of "temp 1 c" in the input data is greater than the threshold value X2, the routine proceeds to the leaf N123, and the execution device 301 diagnoses that the malfunction of the check valve 59 has occurred.

On the other hand, if it is determined at the node N111 that the value of "temp 1 c" in the input data is equal to or less than the threshold value X2, the routine proceeds to the leaf N122, and the execution device 301 makes a diagnosis that the fuel supply system 550 is normal.

When the node N100 determines that the value of "temp 2 d" in the input data is equal to or less than the threshold value X1, the node N110 is entered, and the execution device 301 determines whether or not the value of "time 4 c" in the input data is greater than the threshold value X3. The threshold value X3 is smaller than the threshold value X2.

If the node N110 determines that the value of "time 4 c" in the input data is equal to or less than the threshold value X3, the node N120 is entered, and the execution device 301 diagnoses that the malfunction of the check valve 59 has occurred.

On the other hand, when the value of "time 4 c" in the input data is determined to be greater than the threshold value X3 at the node N110, the node N121 is entered, and the execution device 301 determines whether or not the value of "temp 1 b" in the input data is greater than the threshold value X4. The threshold value X4 is smaller than the threshold value X3.

If it is determined at the node N121 that the value of "temp 1 b" in the input data is greater than the threshold value X4, the routine proceeds to the leaf N131, and the diagnosis is made by the actuator 301 that another abnormality that is not affected by the fuel temperature Tf has occurred.

On the other hand, if it is determined at the node N121 that the value of "temp 1 b" in the input data is equal to or less than the threshold value X4, the routine proceeds to the leaf N130, and the execution device 301 makes a diagnosis that deterioration of the impeller 52c has occurred.

In this way, in the abnormality diagnosis system 600, the diagnosis data acquired by the control device 100 mounted on the vehicle 500 is transmitted to the server device 300 via the communication network 400. The server 300 creates input data using statistical data obtained by counting the data for diagnosis, and diagnoses an abnormality in the fuel supply system 550 using a decision tree, which is a learned model stored in the storage device 302. That is, in the abnormality diagnosis system 600, the control device 100 and the server device 300 connected via the communication network 400 constitute the abnormality diagnosis system 600.

Next, the contents of the routine executed by the control device 100 and the server device 300 in the abnormality diagnosis system 600 to realize the above-described abnormality diagnosis will be described with reference to fig. 6 to 8.

The routine shown in fig. 6 shows a routine related to acquisition of diagnostic data executed by the execution device 101 of the control device 100 mounted on the vehicle 500. This routine is repeatedly executed by the execution device 101 while the fuel pump 52 is operating, on the condition that a predetermined masking time has elapsed since the main switch 140 was turned on and the fuel pump 52 was started. The masking time is set in accordance with the time required for the feed pressure Pf to reach a predetermined level of 300kPa or more in the fuel supply system 550 in a new product state in which no abnormality has occurred.

When the routine is started after the elapse of the masking time, the execution device 101 acquires the feed pressure Pf, which is the fuel pressure in the fuel pipe 57 detected by the fuel pressure sensor 131. Then, in the next step S110, the execution means 101 determines whether the acquired fuel pressure, i.e., the feed pressure Pf, is smaller than the lowest pressure, i.e., the lowest fuel pressure, of the feed pressures Pf acquired in the same stroke. As described later, the minimum fuel pressure is stored in the storage device 102. When the processing of step S110 is executed for the first time after the main switch 140 is turned on, the lowest fuel pressure is not recorded in the storage device 102, and therefore an affirmative determination is made in the processing of step S110.

In the case where it is determined in the process of step S110 that the feed pressure Pf is lower than the minimum fuel pressure (step S110: yes), the execution means 101 advances the process to step S120. Then, the execution means 101 updates the value of the lowest fuel pressure in the process of step S120. That is, the execution device 101 stores the latest value of the feed pressure Pf acquired by the processing of step S100 as the new lowest fuel pressure in the storage device 102.

After the lowest fuel pressure is thus stored, the execution device 101 updates the elapsed time from the activation of the fuel pump 52 in the processing of the next step S130. That is, the execution device 101 stores the elapsed time at this time in the storage device 102 as a new elapsed time.

Next, the execution device 101 advances the process to step S130, and updates the fuel temperature Tf recorded as the diagnostic data. That is, the execution device 101 stores the fuel temperature Tf at this time in the storage device 102 as a new fuel temperature Tf. After the lowest fuel pressure, the elapsed time when the lowest fuel pressure is recorded, and the fuel temperature Tf when the lowest fuel pressure is recorded are stored in the storage device 102 as the data for diagnosis in this way, the execution device 101 once ends the routine.

On the other hand, if it is determined in the process of step S110 that the feed pressure Pf is equal to or higher than the minimum fuel pressure (no in step S110), the execution device 101 temporarily ends the routine without performing the processes of step S120 to step S140.

In this way, in the control device 100, the execution device 101 repeatedly executes the routine, and thereby the minimum fuel pressure during one stroke, the elapsed time when the minimum fuel pressure is recorded, and the fuel temperature Tf are stored as the data for diagnosis in the storage device 102. Then, in the control device 100, when the main switch 140 is turned off and the stroke ends, the execution device 101 causes the transmitter 103 to transmit the diagnostic data stored in the storage device 102.

Upon receiving the data for diagnosis thus transmitted from the transmitter 103 of the control device 100, the execution device 301 of the server device 300 executes the routine shown in fig. 7. When the routine is started, the execution means 301 first updates the first statistical data stored in the storage means 302 in the processing of step S200. As described above with reference to fig. 3, the first statistical data is data obtained by counting the number of times the lowest fuel pressure is recorded, based on the combination of the lowest fuel pressure and the fuel temperature Tf. In the processing of step S200, the execution device 301 increments the number of corresponding regions in the first statistical data by 1 based on the currently received data for diagnosis.

For example, when the lowest fuel pressure value among the received data for diagnosis is 240kPa, the elapsed time is 15 seconds, and the fuel temperature Tf is 35 ℃, the execution means 301 updates the first statistical data by increasing the number of times "temp 2 c" of the first statistical data by 1 in the processing of step S200.

Next, the process proceeds to step S210, and the execution device 301 updates the second statistical data stored in the storage device 302 in the process of step S210. As described above with reference to fig. 4, the second statistical data is data obtained by counting the number of times the lowest fuel pressure is recorded, based on a combination of the lowest fuel pressure and the elapsed time when the lowest fuel pressure is recorded. In the processing of step S210, the execution device 301 increments the number of corresponding regions in the second statistical data by 1 based on the currently received data for diagnosis.

For example, when the lowest fuel pressure value in the received data for diagnosis is 240kPa, the elapsed time is 15 seconds, and the fuel temperature Tf is 35 ℃, the execution means 301 increases the number of times of "time 2 c" of the first statistical data by 1 in the processing of step S210, and updates the second statistical data. When the second statistical data is thus updated, the execution means 301 ends the routine. In the server device 300, the statistical data is updated in this manner every time the diagnostic data is received.

Next, a routine related to the diagnostic process executed by the server device 300 will be described with reference to fig. 8. As described above, the server device 300 performs the diagnosis process every time it receives the data for diagnosis of the fuel supply system 550 to be diagnosed a predetermined number of times.

Specifically, when the predetermined number of times of the diagnosis data is received, the execution device 301 of the server device 300 executes the routine to execute the diagnosis process. When the execution device 301 starts the routine, the values of the times in the first statistical data and the second statistical data are first converted into ratios in the processing of step S300. That is, the execution device 301 creates statistical data converted into the occurrence ratio obtained by dividing the number of occurrences of the lowest fuel pressure held in each region in the statistical data stored in the storage device 302 by the total number of diagnostic data counted up to that time. For example, when the predetermined number of times is 100 times and the total number of diagnostic data counted when the routine is executed is 200 times, 200 occurrences are stored in the first statistical data and the second statistical data, respectively. In this case, the execution device 301 calculates the occurrence ratio, which is a quotient obtained by dividing the number of occurrences stored in each area by "200", and creates statistical data in which the occurrence ratio in each area is stored.

In this way, when the number of occurrences in the statistical data is converted into a ratio, the process proceeds to step S310, and in the process of step S310, the execution device 301 shapes the statistical data stored with the ratio into input data. The input data is data to be input to the learned model stored in the storage device 302 described with reference to fig. 5.

In the processing of step S310, the execution device 301 sets a set of 32 numerical values, which is composed of the values of the respective regions of the first statistical data and the second statistical data converted into the ratios, as 1 input data.

Next, the execution device 301 advances the process to step S320, and inputs the formed input data to the learned model stored in the storage device 302 to execute the diagnosis process. Then, in the processing of the next step S330, the execution device 301 transmits the diagnosis result diagnosed using the learned model to the vehicle 500 mounted with the fuel supply system 550 of the diagnosis target via the transmitter 303. When the diagnosis result is thus transmitted, the execution means 301 terminates the routine.

In this way, in the abnormality diagnosis system 600, the control device 100 mounted on the vehicle 500 includes the storage device 102 that stores the data for diagnosis and the transmitter 103 that transmits the data for diagnosis, and serves as a data transmission device. The server device 300 includes an execution device 301 that performs a diagnostic process, a storage device 302 that stores a learned model, and a receiver 304 that receives diagnostic data, and serves as an abnormality diagnostic device.

When the diagnostic result is received by the receiver 104, the execution device 101 of the control device 100 operates the display unit 150 based on the diagnostic result. If the received diagnosis result indicates that deterioration of the impeller has occurred, the execution device 101 causes the display unit 150 to display information indicating that an abnormality has occurred in the fuel pump 52. On the other hand, when the received diagnosis result indicates that the malfunction of the check valve 59 has occurred, the execution device 101 causes the display unit 150 to display information indicating that an abnormality has occurred in the check valve 59. When the received diagnosis result indicates another abnormality, the execution device 101 displays information indicating that an abnormality has occurred in the fuel supply system. If the diagnosis result indicates normality, the execution device 101 does not particularly operate the display unit 150.

The operation of the present embodiment will be described.

When the fuel pressure in the fuel pipe 57 is lowered while the fuel pump 52 is being operated, there is a possibility that an abnormality occurs in the fuel supply system 550. In the abnormality diagnostic system 600 of the above embodiment, the presence or absence of an abnormality under a specific condition is not diagnosed, but the abnormality is diagnosed based on the lowest fuel pressure and data indicating the state in which the lowest fuel pressure is recorded. Thus, various abnormalities caused by different factors can be detected.

The data indicating the state in which the lowest fuel pressure is recorded is data indicating the state in which the abnormality is estimated to have occurred. Therefore, if the diagnostic data including the data indicating the state in which the minimum fuel pressure is recorded is used, it is possible to estimate the factor that the fuel pressure in the fuel pipe 57 has recorded the minimum fuel pressure.

The effects of the present embodiment will be described.

(1) According to the above-described embodiment in which the abnormality of the fuel supply system 550 is diagnosed using the data for diagnosis, it is possible to not only diagnose the presence or absence of the abnormality, but also determine the location of the failure related to the drop in the fuel pressure in the fuel pipe 57.

(2) The data for diagnosis includes an elapsed time from the start of the fuel pump 52 as data indicating a state in which the lowest fuel pressure is recorded. The elapsed time from the activation of the fuel pump 52 to the recording of the lowest fuel pressure is data indicating the relationship between the activation start time of the fuel pump 52 and the time at which the drop in the fuel pressure occurs. According to the above embodiment, the location of the failure related to the drop in the fuel pressure in the fuel pipe 57 can be determined with reference to whether the drop in the fuel pressure occurs immediately after the start of the fuel pump 52 or when a certain period of time has elapsed after the start.

(3) The check valve 59 is opened when the fuel pump 52 is operating and fuel is being discharged from the fuel pump 52 and there is a flow of fuel from the fuel pump 52 side toward the in-cylinder fuel injection valves 44 and the port fuel injection valves 30 side in the fuel pipe 57. If the check valve 59 is not properly opened due to a malfunction of the check valve 59, a drop in fuel pressure is likely to occur immediately after the fuel pump 52 is started. In contrast, if the impeller 52c of the fuel pump 52 deteriorates, the impeller 52c deforms while the fuel pump 52 is operated, and the impeller 52c interferes with the housing 52b and is hard to rotate. As a result, the fuel pressure is decreased. Such a drop in the fuel pressure due to the deterioration of the impeller 52c is likely to occur in a later period than a period in which a drop in the fuel pressure due to a malfunction of the check valve 59 is likely to occur.

Therefore, when the lowest fuel pressure is recorded after the elapsed time from the start-up of the fuel pump 52 becomes relatively long, the deterioration of the impeller 52c is likely to be a cause of the decrease in the fuel pressure as compared with the malfunction of the check valve 59.

Therefore, as in the above configuration, by using data indicating the relationship between the start time of the fuel pump 52 and the time at which the decrease in the fuel pressure occurs as the diagnostic data, it is possible to determine the deterioration of the impeller 52c of the fuel pump 52 and the malfunction of the check valve 59, and diagnose the abnormality of the fuel supply system 550.

(4) According to the abnormality diagnosis system 600 described above, since the diagnosis is performed using the statistical data obtained by counting the data for diagnosis in a plurality of trips, it is possible to perform diagnosis with higher accuracy than the case where the diagnosis is performed based on the data for diagnosis in only one trip.

(5) The fuel temperature Tf at which the lowest fuel pressure is recorded is data representing the relationship of the fuel temperature Tf and the drop in fuel pressure. When an abnormality occurs in the fuel supply system 550 and the fuel pressure decreases, the degree of influence of the temperature of the fuel on the decrease in the fuel pressure differs depending on the location where the failure occurs. According to the abnormality diagnostic system 600 described above, it is possible to determine the location of a failure related to a drop in fuel pressure in the fuel pipe 57 with reference to whether the drop in fuel pressure occurs when the temperature of the fuel is low or when the temperature of the fuel is high.

(6) When the impeller 52c deteriorates, the impeller 52c deforms with an increase in the temperature of the fuel and the temperature of the impeller 52c, and the impeller 52c interferes with the casing 52b, so that the fuel pressure is likely to decrease. Further, the check valve 59 is more likely to cause malfunction as the temperature of the fuel is lower. Therefore, by performing diagnosis using statistical data obtained by counting the diagnostic data including the data in which the temperature of the fuel at the time of the lowest fuel pressure is recorded as described above, it is possible to diagnose whether the drop in the fuel pressure is deterioration of the impeller 52c affected by the temperature of the fuel, malfunction of the check valve 59, or other abnormality not affected by the temperature of the fuel.

(7) It is possible to create a model for diagnosing the presence or absence of an abnormality from statistical data and outputting a category of a factor of the abnormality when the abnormality occurs, using machine learning. Using machine learning, it is possible to extract features that are difficult for a person to perceive and perform abnormality diagnosis. In addition, if a learned model in which statistical data obtained by counting diagnostic data is input is used as in the abnormality diagnosis system 600 described above, the amount of input data can be reduced compared to a case where a learned model in which a plurality of diagnostic data themselves are input is constructed.

(8) Compared with models such as a neural network and the like, the decision tree is easy for people to understand the basis of diagnosis of the learned models. According to the above-described abnormality diagnosis system 600, it is possible to construct the abnormality diagnosis system 600 in which the reason for deriving the diagnosis result can be easily explained.

(9) The abnormality diagnosis system 600 includes, as storage devices, a storage device 102 mounted on the vehicle 500 and storing data for diagnosis, and a storage device 302 mounted on a server device 300 which is a device different from the vehicle 500 and storing a learned model. That is, the abnormality diagnosis system 600 includes the storage device 102 as the first storage device mounted on the vehicle 500 side and the storage device 302 as the second storage device mounted on the server device 300 side.

The execution device 301 mounted on the server device 300 together with the storage device 302 receives the diagnostic data stored in the storage device 102 from the vehicle 500 and creates statistical data. The execution device 301 uses the created statistical data to determine the cause of an abnormality related to a drop in the fuel pressure in the fuel pipe 57 using the learned model stored in the storage device 302, thereby diagnosing an abnormality in the fuel supply system 550.

With this configuration, the server device 300 different from the vehicle 500 performs the generation of the statistical data, the diagnosis of the abnormality by the learned model, and the like. Therefore, an increase in the capacity of the storage device 302 on the vehicle 500 side and an increase in the calculation load on the vehicle 500 side can be suppressed.

This embodiment can be modified and implemented as follows. This embodiment and the following modifications can be combined and implemented within a range not technically contradictory to the technology.

In the above-described abnormality diagnosis system 600, the example in which the learned model is a decision tree is illustrated, but the learned model used in the diagnosis process does not necessarily have to be a decision tree. For example, the learned model used in the diagnosis process may be a random forest in which the diagnosis result is determined by majority determination of a plurality of decision trees. The learned model used in the diagnostic process may be a neural network as shown in fig. 9.

In the example shown in fig. 9, an input layer is provided that is configured by 32 nodes (node N01 to node N32) and the 32 nodes input a value obtained by converting the number of occurrences in each area in the first statistical data into a proportional value and a value obtained by converting the number of occurrences in each area in the second statistical data into a proportional value. The neural network includes an intermediate layer including 4 nodes (node N41 to node N44) and an output layer including 4 nodes (node N51 to node N54).

In the neural network shown in fig. 9, the activation function of the intermediate layer is a sigmoid function. The input to the intermediate layer is calculated as the sum of values obtained by multiplying each of the 32 input values to the input layer by a weight. Then, the sum of the values obtained by multiplying the output values of the nodes (node N41 to node N44) in the middle layer by the weights is input to the output layer. The input values to these output layers are input to the output layer as the compliance maximum layer, and are converted into output values corresponding to the nodes (node N51 to node N54). The sum of the output values of the nodes (node N51 to node N54) of the output layer is "1", and each output value represents a ratio to "1". The nodes (node N51 to node N54) of the output layer correspond to the diagnostic results output by the diagnostic process. The types of diagnosis results are the same as those in the above embodiment. For example, the node N51 corresponds to "normal", and the node N52 corresponds to "deterioration of the impeller". The node N53 corresponds to "malfunction of the check valve 59", and the node N54 corresponds to "other abnormality". That is, in the output layer, the probabilities corresponding to the 4 diagnosis results of "normal", "deterioration of impeller", "operation failure of check valve 59", and "other abnormality" are output.

When supervised learning is performed by inputting teacher data similar to that of the above-described embodiment to such a neural network, a learning completion model capable of diagnosing an abnormality of the fuel supply system 550 similar to that of the above-described embodiment by the same input as that of the above-described embodiment can be generated.

The neural network shown in fig. 9 includes only 1-layer intermediate layers, but the number of intermediate layers may be any number of 2 or more layers, and the number of nodes in the intermediate layers may also be any number.

In the above-described embodiment, the example in which the abnormality diagnosis is performed using the learned model obtained by the machine learning has been described, but the learned model obtained by the machine learning does not necessarily need to be used. For example, it is also possible to find the threshold of the branch by repeating experiments and verifications and to construct a model such as the decision tree of the above embodiment.

In the above-described embodiment, the example in which the abnormality diagnosis is performed 1 time based on the statistical data obtained by counting the data of the predetermined number of strokes has been described, but the abnormality diagnosis may be performed for each stroke based on the lowest fuel pressure and the information obtained when the lowest fuel pressure is recorded.

Although the example in which the fuel temperature Tf at the time when the lowest fuel pressure is recorded is included in the data for diagnosis is shown, a configuration in which the fuel temperature Tf is not included in the data for diagnosis and abnormality diagnosis is performed may be employed.

For example, the decision tree shown in fig. 10 is an example of a decision tree for performing diagnostic processing based on only the second statistical data. The decision tree can be generated by supervised learning using teacher data composed of 17 values of the second statistical data and the label of the positive solution. Note that, since information on the fuel temperature Tf is not included in the second statistical data, the presence or absence of the influence of the fuel temperature Tf is not considered in the decision tree. Therefore, in the output of this decision tree, an output corresponding to "other abnormality" in the above-described embodiment is not performed. That is, the label in the teacher data is 3 types of "normal", "impeller", and "check valve", and the diagnosis result output by the decision tree is 3 types of "normal", "deterioration of impeller", and "malfunction of check valve".

As shown in fig. 10, in the decision tree, first, it is determined by the execution means 301 at the node N200 whether or not "time 1 c" in the input data is larger than the threshold Y1. If "time 1 c" in the input data is equal to or less than the threshold value Y1, the routine proceeds to the leaf N210, and the execution device 301 makes a normal diagnosis.

On the other hand, if "time 1 c" in the input data is greater than the threshold value Y1, the node N211 is entered, and the execution device 301 determines whether "time 4 c" in the input data is greater than the threshold value Y2. The threshold value Y2 is a value smaller than the threshold value Y1.

If "time 4 c" in the input data is greater than the threshold value Y2, the routine proceeds to the leaf N221, and the execution device 301 makes a diagnosis that deterioration of the impeller 52c has occurred. On the other hand, if "time 4 c" in the input data is equal to or less than the threshold value Y2, the routine proceeds to the leaf N220, and the execution device 301 diagnoses that malfunction of the check valve 59 has occurred.

The content of the data for diagnosis is not limited to the above example. For example, diagnostic data that does not include information about the elapsed time at which the lowest fuel pressure is recorded may be used. The information indicating the state in which the lowest fuel pressure is recorded, which is included in the diagnostic data, may not be the fuel temperature Tf or the elapsed time.

It is also possible to provide: the control device 100 as a data transmission device performs statistics of data for diagnosis, and the control device 100 transmits the statistical data to the server device 300 via the transmitter 103. In this case, the server device 300 may create input data from the received statistical data and perform the diagnosis process.

The abnormality diagnosis system may be completed only in control device 100 mounted on vehicle 500. That is, it is also possible to: the learned model is stored in the storage device 102, and the control device 100 executes processes from acquisition of the data for diagnosis to statistics, generation of input data, and diagnosis. Further, it is also possible to: the diagnostic data is transmitted from the control device 100 as in the above-described embodiment, but only the statistical data is generated in the server device 300, the statistical data is transmitted from the server device 300 to the control device 100, and the diagnostic process is performed in the control device 100.

Although the control device 100 controls the fuel pump 52 by the fuel pump control device 200, a configuration may be adopted in which 1 control device having the functions of both the control device 100 and the fuel pump control device 200 is provided. The control device of the fuel supply system 550 may be constituted by 3 or more units.

In the above embodiment, the occurrence of an abnormality is reported by visual information by operating the display unit 150, but the present invention is not limited to this. For example, it is also possible to report the occurrence of an abnormality by audible information by operating a speaker.

The execution device 101 and the execution device 301 are not limited to executing software processing. For example, a dedicated hardware circuit (e.g., ASIC) may be provided for performing hardware processing on at least a part of the software processing in the above embodiment. That is, the execution device 101 and the execution device 301 may have any of the following configurations (a) to (c). (a) The processing device is provided with a processing device for executing all the above-mentioned processing according to a program, and a program storage device such as a ROM for storing the program. (b) The apparatus includes a processing device and a program storage device for executing a part of the above-described processing in accordance with a program, and a dedicated hardware circuit for executing the remaining processing. (c) The apparatus includes a dedicated hardware circuit for executing all of the above-described processing. Here, the software executing apparatus and the dedicated hardware circuit provided with the processing apparatus and the program storage apparatus may be plural ones.

In the above-described embodiment, the fuel supply system 550 including the in-cylinder fuel injection valve 44 and the port fuel injection valve 30 as the fuel injection valves is exemplified, but the configuration of the fuel supply system is not limited to such a configuration. For example, a fuel supply system including only port fuel injection valves may be used. For example, a fuel supply system including only in-cylinder fuel injection valves may be used.

The vehicle 500 is not limited to a vehicle in which the device for generating the vehicle propulsion is the engine only, and may be a series hybrid vehicle, for example. In addition, a parallel hybrid vehicle or a hybrid vehicle may be used in addition to the series hybrid vehicle.

Although the example in which the fuel temperature Tf is detected by the fuel temperature sensor 137 is shown, the fuel temperature Tf may be obtained by estimation.