阅读说明：本技术 一种航空活塞发动机增压自适应控制方法及系统 (Aviation piston engine supercharging self-adaptive control method and system ) 是由 姜梁 董雪飞 王振宇 陈建国 司亮 刘培强 于 2021-06-28 设计创作，主要内容包括：本发明公开了一种二冲程航空活塞增压发动机自适应控制方法,属于二冲程航空活塞增压发动机领域。该方法包括：任务时序规划步骤：根据发动机控制需求进行任务时序规划；工作状态评估步骤：采集工作参数及接收指令并评估发动机工作状态；控制参数自学习及预设步骤：根据历史运行数据通过神经网络算法进行自学习并预设控制参数；故障识别/诊断/响应步骤：实时监测发动机部件状态,针对故障问题进行识别、诊断与响应。该方法是一种具有一定适应能力的控制方法,能够自动适时地调节发动机控制参数,以适应外界环境、运行工况变化的影响,使发动机在不同环境及工况下始终维持最优或次优状态,满足系统使用要求。(The invention discloses a self-adaptive control method for a two-stroke aviation piston supercharged engine, and belongs to the field of two-stroke aviation piston supercharged engines. The method comprises the following steps: and a task time sequence planning step: performing task time sequence planning according to the control requirement of the engine; and a working state evaluation step: collecting working parameters, receiving instructions and evaluating the working state of an engine; controlling parameters to self-learn and preset: self-learning is carried out through a neural network algorithm according to historical operation data and control parameters are preset; fault identification/diagnosis/response step: and the states of the engine components are monitored in real time, and the fault problem is identified, diagnosed and responded. The method is a control method with certain adaptability, can automatically and timely adjust the control parameters of the engine to adapt to the influence of the external environment and the change of the operating condition, so that the engine can always maintain the optimal or suboptimal state under different environments and working conditions, and the use requirement of the system is met.)

1. A self-adaptive control method for a two-stroke aviation piston supercharged engine is characterized in that changes of environment and engine running states can be automatically identified, engine control parameters can be adjusted in real time, and the engine can always maintain an optimal or suboptimal running state under the current working condition, and the method comprises the following steps:

and a task time sequence planning step: performing task time sequence planning according to the engine control task;

and a working state evaluation step: collecting working parameters, receiving instructions and evaluating the working state of an engine;

controlling parameters to self-learn and preset: self-learning is carried out through a neural network algorithm according to historical operation test data and control parameters are preset;

fault identification/diagnosis/response step: and the states of the engine components are monitored in real time, and the fault problem is identified, diagnosed and responded.

2. The adaptive control method for a two-stroke aviation piston supercharged engine of claim 1, wherein the mission schedule planning step specifically comprises:

s11: the engine control task is in modular design, and components are divided according to the control task to generate a modular control task;

s12: and dividing the time sequence and the priority of the modularized control task.

3. The adaptive control method for a two-stroke aviation piston supercharged engine of claim 1, wherein said operating condition assessment step comprises:

s21: calculating corresponding parameter physical values of the engine according to working parameters acquired by the sensor, wherein the corresponding parameter physical values comprise throttle opening, rotating speed, cylinder head temperature, atmospheric temperature and altitude;

s22: judging the working conditions of the engine, including a starting working condition, a transition working condition, a warming working condition and a steady-state working condition, according to the rotating speed and the cylinder head temperature;

s23: judging the working gears of the engine according to the opening degree and the rotating speed of a throttle valve, wherein the working gears comprise a slow gear, an idle gear, a cruise gear and a big gear;

s24: and judging the working environment of the engine according to the altitude and the atmospheric temperature.

4. The adaptive control method for the two-stroke aviation piston supercharged engine according to claim 1, wherein the self-learning and presetting steps of the control parameters specifically comprise:

s31: transmitting and storing correction coefficient data of atmospheric pressure, an air door, rotating speed, cylinder head temperature, atmospheric temperature and an oil way in real time in the process of engine flight and bench test as historical operation test data;

s32: based on historical operation test data, performing initial training on the neural network, performing iterative computation on the neural network in real time through current operation data in the operation process of the engine, and continuously optimizing the neural network coefficient by using the latest data;

s33: in the running process of the engine, reading the altitude, the opening of a throttle valve, the rotating speed, the cylinder head temperature and the atmospheric temperature by using the trained neural network, and calculating an oil injection correction coefficient as a preset control parameter;

s34: and on the basis of the preset control parameters, correcting the basic fuel injection quantity by adopting a closed-loop correction algorithm based on the air-fuel ratio.

5. The adaptive control method for the two-stroke aviation piston supercharged engine according to claim 4, characterized in that the neural network adopts a local linear model tree model developed based on a radial basis function neural network, and replaces the weight coefficients of the output layer of the radial basis function neural network with linear functions.

6. The adaptive control method for the two-stroke aviation piston supercharged engine according to claim 5, characterized in that the neural network is divided into 6 neurons according to altitude.

7. The adaptive control method for a two-stroke aviation piston supercharged engine of claim 1, wherein said fault identification/diagnostic/response step comprises in particular:

and monitoring sensor data in real time, diagnosing and judging whether the sensor data is abnormal according to sensor signals, monitoring the state of engine parts, and timely adopting dimension reduction control or redundancy control when the sensor fails.

8. The adaptive control method for the two-stroke aviation piston supercharged engine of claim 7, wherein the dimension reduction control specifically comprises:

monitoring preset control parameters in real time through a sensor fault diagnosis algorithm, canceling preset control when the preset control parameters are abnormal, and adopting closed-loop control based on an air-fuel ratio; when the air-fuel ratio data is abnormal, the preset control and the closed-loop control are directly cancelled, and an open-loop control mode is adopted.

9. The adaptive control method for a two-stroke aviation piston supercharged engine of claim 7, wherein said redundant control comprises:

the sensor fault diagnosis algorithm is used for monitoring two paths of sensors respectively contained in the air door position sensor and the rotating speed sensor in real time, and the sensor data are switched to the standby sensor immediately when abnormal conditions occur.

10. An adaptive control system for a two-stroke aviation piston supercharged engine, wherein the system is adapted to control based on the method of any one of claims 1 to 9, the system comprising:

the task time sequence planning component is used for planning the task time sequence according to the engine control requirement;

the working state evaluation component is used for acquiring working parameters, receiving instructions and evaluating the working state of the engine;

the control parameter self-learning and presetting component is used for carrying out self-learning through a neural network algorithm according to historical operating data and presetting control parameters;

and the fault identification/diagnosis/response component is used for monitoring the state of the engine component in real time and identifying, diagnosing and responding aiming at fault problems.

Technical Field

The invention is applied to the field of two-stroke aviation piston supercharged engines, in particular to an aviation piston two-stroke supercharged engine self-adaptive control method and system, and is applied to two-stroke aviation piston supercharged engines.

Background

Compared with the four-stroke piston engine supercharging technology, the two-stroke piston engine introduces the supercharging technology due to the strong coupling of the internal flow field in the air exchange process, and needs to effectively solve the key technologies such as the accurate control of the combustion of the mixed gas and the like. The supercharged engine needs to recycle exhaust pulse energy to do work, and the supercharging system is coupled with the air exchange process, so that the fluctuation of fresh air sealed in a cylinder under the same air door and rotating speed is large, and the oil injection quantity cannot be accurately controlled. The operating condition of the engine is complicated and changeable under the full-flight envelope, and the conventional control method is adopted, so that the fuel injection amount control is difficult to respond to the sudden change of the air input of the engine along with the change of the working condition of a supercharging system in real time, the proportion of fuel and air is not matched, the problem of unstable combustion is caused, and the failure of the engine is easily caused.

The stable combustion is the core of engine control, and the stable power output is ensured by controlling the air-fuel ratio in the cylinder in a proper interval. When ground calibration is carried out, the altitude can only be below 5000m, but the engine needs to work at a higher altitude, so that in order to ensure stable combustion of the engine and maintain the robustness of system control, a control strategy is required to realize accurate control on the engine under different working scenes, and the engine can exert power output to the maximum extent.

Disclosure of Invention

The invention aims to overcome the defects of the prior art and provides a task planning-based adaptive control method and system for a two-stroke aviation piston supercharged engine, wherein the method comprises the technologies of time sequence control, neural network self-learning control, pre-control + closed-loop oil circuit control technology, fault monitoring, identification and response and the like, is a control method with certain adaptive capacity, can automatically and timely adjust engine control parameters to adapt to the influence of external environment and operation condition change, enables the engine to always maintain an optimal or suboptimal state under different environments and working conditions, and meets the use requirements of the system.

According to a first aspect of the embodiments of the present invention, there is provided an adaptive control method for a two-stroke aviation piston supercharging engine, the method comprising:

and a task time sequence planning step: performing task time sequence planning according to the engine control task;

and a working state evaluation step: collecting working parameters, receiving instructions and evaluating the working state of an engine;

controlling parameters to self-learn and preset: self-learning is carried out through a neural network algorithm according to historical operation test data and control parameters are preset;

fault identification/diagnosis/response step: and the states of the engine components are monitored in real time, and the fault problem is identified, diagnosed and responded.

Further, the task timing planning step specifically includes:

s11: the engine control task is in modular design, and components are divided according to the control task to generate a modular control task;

s12: and dividing the time sequence and the priority of the modularized control task.

Further, the S11 specifically includes: and arranging the engine control tasks, adopting a modular design, and dividing the components aiming at the engine control tasks to generate the modular control tasks.

Further, the component divisions include, but are not limited to, engine start/stop control, air intake system, fuel injection system, ignition system, exhaust system, fault diagnosis, basic functional components.

Further, the S12 specifically includes: designing and planning the modularized control tasks, and dividing the modularized control tasks of the engine into three types from high to low according to real-time requirements and priorities except for an initialization task: interrupt handling tasks, periodic handling tasks, background tasks.

Further, the interrupt handling task priority comprises: the method comprises the following steps of rotating speed acquisition, oil injection control and ignition control tasks, wherein once an interruption processing task is triggered, the current task is suspended, and the interruption processing task is continued after the execution of the interruption processing task is finished.

Further, the periodic processing task includes: the system comprises an MCU (microprogrammed control Unit) initialization task, a software initialization task, a hardware initialization task, a sensor signal acquisition task, a sensor signal processing task, a working condition judgment task, an actuator driving task and a communication task.

Further, the periodic processing task is divided into periodic tasks of 5ms, 10ms and 15ms according to requirements.

Further, the background task is a task executed when the uninterrupted or periodic task runs, and is used for continuously calculating the rotating speed.

Further, the working state evaluating step specifically includes:

s21: calculating corresponding parameter physical values of the engine according to working parameters acquired by the sensor, wherein the corresponding parameter physical values comprise throttle opening, rotating speed, cylinder head temperature, atmospheric temperature and altitude;

s22: judging the working conditions of the engine, including a starting working condition, a transition working condition, a warming working condition and a steady-state working condition, according to the rotating speed and the cylinder head temperature;

s23: judging the working gears of the engine according to the opening degree and the rotating speed of a throttle valve, wherein the working gears comprise a slow gear, an idle gear, a cruise gear and a big gear;

s24: and judging the working environment of the engine according to the altitude and the atmospheric temperature.

Further, the control parameter self-learning and presetting step specifically comprises the following steps:

s31: transmitting and storing correction coefficient data of atmospheric pressure, an air door, rotating speed, cylinder head temperature, atmospheric temperature and an oil way in real time in the process of engine flight and bench test as historical operation test data;

s32: based on historical operation test data, performing initial training on the neural network, performing iterative computation on the neural network in real time through current operation data in the operation process of the engine, and continuously optimizing the neural network coefficient by using the latest data;

s33: in the running process of the engine, reading the altitude, the opening of a throttle valve, the rotating speed, the cylinder head temperature and the atmospheric temperature by using the trained neural network, and calculating an oil injection correction coefficient as a preset control parameter;

s34: and on the basis of the preset control parameters, correcting the basic fuel injection quantity by adopting a closed-loop correction algorithm based on the air-fuel ratio.

Further, the neural network adopts a Local Linear Model Tree (Local Linear Model Tree) Model expanded based on a Radial Basis Function (RBF) neural network, and replaces the weight coefficient of the output layer of the RBF neural network with a Linear Function. Therefore, the neural network can map any complex nonlinear relation, has strong robustness, memory capability, nonlinear mapping capability and self-learning capability, and can accurately fit the nonlinear relation between the fuel injection quantity and the altitude, the throttle opening, the rotating speed and the temperature. And the radial basis function neural network learning rule is simple, and effective operation on real-time data can be realized.

Further, in the neural network, the neurons are divided according to the altitude, and the total number of the neurons is 6. The altitude of the current system is determined through atmospheric pressure, data of each altitude is used for learning of one neuron, the neural network is trained through engine operation data, and the weight of each neuron under different altitudes is obtained.

Further, during the operation of the engine, the corresponding weight of the neuron is far higher than that of the rest neurons according to different altitudes.

Further, the fault identifying/diagnosing/responding step specifically includes:

and monitoring the sensor data in real time, diagnosing and judging whether the sensor data is abnormal according to the sensor signal, and timely adopting dimension reduction control or redundancy control when the sensor fails.

Further, the sensor signal diagnosing includes:

the method comprises the steps that drive stage diagnosis is conducted, aiming at a complete failure fault of a sensor, signals under the condition of a line fault of the sensor are obtained through a short power supply, a short ground and an open circuit mode under the condition that an engine is not in a running state, the signal range is determined when the sensor works normally, software is written in the signal range, after an Electronic Control Unit (ECU) receives signals of the sensor, in order to prevent fault misjudgment, average filtering is conducted on 10 signals collected in each group, and whether the filtered value is within the normal working signal range is judged;

and (3) diagnosing rationality, namely, aiming at the sensor precision reduction fault, determining the critical value through a critical value judgment method, wherein the critical value is determined through the maximum value and the minimum value of the filtered numerical value under the same working condition, calculating the maximum value and the minimum value once again after receiving one datum, calculating the weighted average value of the maximum value and the minimum value to obtain a critical range, and the critical numerical value is always between the maximum value and the minimum value of the existing sensor actual signal.

Further, the dimension reduction control specifically includes:

monitoring preset control parameters in real time through a sensor fault diagnosis algorithm, canceling preset control when the preset control parameters are abnormal, and adopting closed-loop control based on an air-fuel ratio; when the air-fuel ratio data is abnormal, the preset control and the closed-loop control are directly cancelled, and an open-loop control mode is adopted.

Further, the redundancy control specifically includes:

the sensor fault diagnosis algorithm is used for monitoring two paths of sensors respectively contained in the air door position sensor and the rotating speed sensor in real time, and the sensor data are switched to the standby sensor immediately when abnormal conditions occur.

According to a second aspect of embodiments of the present invention, there is provided an adaptive control system for a two-stroke aviation piston supercharged engine, the system comprising:

the task time sequence planning component is used for planning the task time sequence according to the engine control requirement;

the working state evaluation component is used for acquiring working parameters, receiving instructions and evaluating the working state of the engine;

the control parameter self-learning and presetting component is used for carrying out self-learning through a neural network algorithm according to historical operating data and presetting control parameters;

and the fault identification/diagnosis/response component is used for monitoring the state of the engine component in real time and identifying, diagnosing and responding aiming at fault problems.

The invention has the beneficial effects that:

1. the self-adaptive control method carries out task planning based on the control requirement of the engine, carries out frequency division aiming at the precision requirement and the real-time requirement of different tasks, and effectively improves the software operation efficiency and the control precision of key parameters of the engine;

2. the self-adaptive control method adopts a self-learning and closed-loop correction method, a self-learning strategy learns engine historical data through a neural network based on a radial basis function and corrects the historical data through real-time operation data to obtain an optimal oil injection control parameter of the engine under real-time working conditions, and closed-loop correction is performed on the basis of the control parameter, so that closed-loop correction time can be effectively reduced, the control response speed of the engine is improved, and the engine can keep the optimal working performance under each working condition;

3. the self-adaptive control method monitors all parts of the engine in real time according to the running state of the engine and adopts a redundancy control or dimension reduction control strategy for partial faults so as to ensure that the engine can keep stable running after the faults occur.

Drawings

FIG. 1 illustrates a system composition diagram of the present invention;

FIG. 2 illustrates a self-learning algorithm framework diagram of the present invention.

Detailed Description

As shown in fig. 1, the adaptive control method of the present invention is applicable to an engine EMS system including an engine controller ECU, system components, and a wire harness. The EMS system of the engine is a key factor for ensuring the normal work of the engine, and needs to reasonably design the components and the system composition and the reliable electronic injection control of the engine, ensure the reliable power output and simultaneously consider the goals of functional safety and lowest oil consumption. The ECU reasonably controls the processes of air intake, oil injection and ignition by collecting signals of the crankshaft position, the air intake temperature, the air intake pressure, the throttle valve position and the like, so that the engine works in the optimal state.

The basic functions of the self-adaptive control method are to realize the acquisition, calculation and actuator driving actions of signals, sweep the mixed oil gas into a cylinder at a specific angle window, ignite, burn and do work, and ensure the stable combustion of the engine. In addition, in order to ensure the reliability of engine control, the functional safety design, the fault diagnosis and the fault post-processing function are required to be performed at the same time, and the method specifically includes:

task timing planning

The time sequence control technology for planning tasks according to the urgency degree and the timeliness divides the priority and the frequency of the tasks according to the importance, and adopts different execution frequencies according to the difference of the real-time requirements of the tasks. The method comprises the following steps:

s1: the engine control task is in modular design, and components are divided according to the control task;

s2: and dividing the time sequence and the priority of the modularized control task.

The specific method of the time sequence control technology comprises the following steps:

1) the method comprises the following steps of arranging engine control tasks, adopting a modular design, and carrying out component division aiming at the engine control tasks, wherein the component division comprises components such as engine starting/stopping control, an air inlet system, an oil injection system, an ignition system, an exhaust system, fault diagnosis, basic functions and the like;

2) designing and planning the control tasks, and dividing the engine control tasks into three types according to real-time requirements and emergency degrees except for an initialization task: interrupt handling tasks, periodic handling tasks, background tasks. The priority of the interrupt processing task is highest and comprises a rotating speed acquisition task, an oil injection control task and an ignition control task, once the interrupt processing task is triggered, the current task is suspended, and the interrupt processing task is continued after the execution of the interrupt processing task is finished. The periodic processing tasks comprise tasks of cylinder temperature and other data acquisition, partial control parameter calculation and the like, and are divided into periodic tasks of 5ms, 10ms, 15ms and the like according to requirements. The background task is a task executed when the uninterrupted or periodic task runs and is used for continuously calculating the rotating speed.

Working state assessment

Control parameter self-learning and presetting

In order to solve the problems that closed-loop calculation is slow in convergence caused by large closed-loop correction amount and experimental calibration cannot be carried out under high-altitude operation conditions, a pre-control and closed-loop control technology based on neural network self-learning is provided, and the method comprises the following steps:

s1: parameters such as altitude, air door, rotating speed, temperature and the like and correction coefficient data of an oil way are transmitted to a storage chip in real time and stored in the process of engine flight and bench test;

s2: based on a large amount of experimental data, the neural network is initially trained. Meanwhile, in the running process of the engine, iterative calculation is carried out on the neural network through running data in real time, and the neural network coefficient is optimized by continuously utilizing the latest data.

S3: in the running process of the engine, reading parameters such as altitude and the like by using the trained neural network, and calculating an oil injection correction coefficient as a pre-control parameter;

s4: and on the basis of pre-control, correcting the basic fuel injection quantity by adopting a closed-loop correction algorithm based on the air-fuel ratio.

The neural network adopts a LOLIMOT model obtained based on RBF expansion, and the weight coefficient of the output layer of the RBF network is replaced by a linear function. In the neural network, the neurons are divided according to the altitude, and the total number of the neurons is 6.

The neural network self-learning-based pre-control strategy is characterized in that an oil way self-learning strategy based on a neural network is added on the basis of the existing closed-loop control. The specific implementation mode of the self-learning strategy is as follows:

s1: transmitting and storing the data of atmospheric pressure, air door, rotating speed, cylinder head temperature, atmospheric temperature and correction coefficient of an oil way to a storage chip in real time in the process of engine flight and bench test;

s2: based on a large amount of experimental data, the neural network is initially trained. Meanwhile, in the running process of the engine, iterative calculation is carried out on the neural network through running data in real time, and the neural network coefficient is optimized by continuously utilizing the latest data.

S3: in the running process of the engine, reading parameters such as atmospheric pressure and the like by using the trained neural network, and calculating an oil injection correction coefficient as a pre-control parameter;

s4: and on the basis of pre-control, correcting the basic fuel injection quantity by adopting a closed-loop correction algorithm based on the air-fuel ratio.

The neural network adopts a LOLIMOT model obtained based on RBF expansion, and the weight coefficient of the output layer of the RBF network is obtained by replacing a linear function. As shown in FIG. 2, the LOLIMOT model contains a plurality of neurons, and unlike other neural network structures, each neuron of the LOLIMOT model represents a local sub-module, and each local sub-module contains a high-dimensional Gaussian function lambda_iAnd a linear function LLM_mThe output of the local submodel is the Gaussian function output phi_iAnd linear function outputProduct of (d), LOLIMOT model Total outputThe sum of the normalized outputs of the local models is used.

Let x be ═ x₁,x₂,…,x_p]Setting the total number of local submodels in the LOLIMOT network as M for the p-dimensional input vector of the model, and setting the Gaussian effective function phi of the mth (M is more than or equal to 1 and less than or equal to M) local submodel_mThe output is:

wherein, c_m,iIs the center of the Gaussian function of the ith (i is more than or equal to 1 and less than or equal to p) dimensional data of the mth local sub-model_m,iGaussian function for ith dimension data of mth submodelNumber variance, μ_m(x) Is the output value, μ, of the m-th local submodel_j(x) Is the output value of the j (j is more than or equal to 1 and less than or equal to M) th local sub-model. Linear function LLM of mth local submodel_mOutput is as

Wherein, ω is_m,iCoefficient of linear equation, mu, for the ith dimension data of the mth local submodel_m,iThe weight coefficient of the m local sub-model Gaussian function output is output, and the output of the LOLIMOT model is

In the above formula, the updating of the gaussian function center in the LOLIMOT submodel is performed in a data axial division manner, and the linear parameter ω of the local model is_m,iAnd adopting a weighted least square method for estimation.

The neurons of the invention are divided and trained according to the altitude, and the data of the altitude of every 1000m is used for training and learning one neuron. In the operation process of the engine, the corresponding weight of the neuron is far higher than that of other neurons according to different altitudes. During the operation of the engine, the neural network reads parameters such as atmospheric pressure and the like, calculates to obtain pre-control parameters of the oil injection pulse width, and then corrects the pre-control parameters through a closed-loop control algorithm based on the air-fuel ratio to obtain final control parameters.

The range of the operating condition of the engine is wide, the required air-fuel ratios of the mixed gas are different under different rotating speeds, in order to ensure that the engine can meet the requirements of dynamic property and reliability under different operating condition modes, an air-fuel ratio zone control mode is adopted, different rotating speed regions are controlled based on different target air-fuel ratios, and the target air-fuel ratios in the different rotating speed regions are shown in the following table 1.

TABLE 1 target air-fuel ratio for different speed intervals

Interval of rotation/rpm	Air-fuel ratio demand
		Less than 3000	8.9-9.1
3000-3600	8.4-8.8
		3600-3850	8.6-8.8
3850-4000	8.8
		4000-5000	9.1
5000-5200	8.9
		5200 or more	8.5

PID closed-loop control is adopted, the closed-loop correction amount is proportional term + integral term + differential term, the proportional correction term is a reverse correction amount applied when the concentration change of the mixed gas is detected, when the mixed gas is lean, the mixed gas is enriched by a certain step length, and when the mixed gas is rich, the gas injection amount is reduced by a certain step length, so that the mixed gas is always kept near the target air-fuel ratio. The integral term compensates the deviation of the actual air-fuel ratio to the target air-fuel ratio, and influences the fluctuation and amplitude of the air-fuel ratio, the longer the integral step length is, the shorter the calculation period is, and the faster the integral correction is, so the small step length is used for medium and small loads, and the large step length is used for large loads, and in addition, in order to ensure that the integral correction term is in a reasonable range, the integral correction amount needs to be limited, the limit value represents the correction amount of the oil injection pulse width, the limit of the medium and small loads is smaller, and the limit of the large load is larger. After being calibrated by a plurality of tests, the PID control parameter can quickly adjust the oil injection pulse width to stabilize the air-fuel ratio near the target air-fuel ratio.

Fault identification/diagnosis/response

According to data of each sensor of the engine, a fault diagnosis and response technology comprising fault detection, fault identification, fault diagnosis and fault response is provided, the data of the sensors are monitored in real time, a redundancy control strategy or a dimension reduction control strategy is switched according to the fault type, and the stable operation of the engine is guaranteed when partial devices are in fault.

The fault diagnosis and response technology monitors the sensor data in real time, judges whether the sensor data is abnormal or not, and adopts dimension reduction control or redundancy control in time when the sensor fails.

The sensor signal diagnostics are classified as drive level diagnostics and signal rationality diagnostics. The method comprises the steps that for a complete failure fault of a sensor, in the non-running state of an engine, signals under the condition of line faults of the sensor are obtained through a short power supply, a short ground and an open circuit mode, the signal range of the sensor in normal working is determined and written into software, after an ECU receives signals of the sensor, in order to prevent fault misjudgment, average filtering is carried out on 10 signals collected every time, and whether the filtered values are in the normal working signal range or not is judged. And aiming at the sensor precision reduction fault, the rationality diagnosis also determines the critical value through a critical value judgment method, the critical value is determined through the maximum value and the minimum value of the filtered numerical value under the same working condition, the maximum value and the minimum value are recalculated once every time one datum is received, the weighted average value of the maximum value and the minimum value is calculated to obtain a critical range, and the critical numerical value is always between the maximum value and the minimum value of the appeared actual signal of the sensor.

The fault diagnosis response technology comprises the following steps of dimension reduction control and redundancy control:

the opening degree and the rotating speed of the air door are main parameters, and redundancy control is adopted. The air door position sensor and the rotating speed sensor respectively comprise two paths of sensors, namely a main sensor and a standby sensor, real-time monitoring is carried out through a sensor fault diagnosis algorithm, and the sensor data are immediately switched to the standby sensor when abnormal conditions occur.

And the ECU oil circuit control adopts dimension reduction control. In the self-adaptive control method, the oil circuit control adopts a pre-control and closed-loop control mode based on neural network self-learning, wherein the pre-control requires data such as altitude, an air door, rotating speed, cylinder temperature and the like, the required parameters are monitored in real time through a sensor fault diagnosis algorithm, when the data are abnormal, the pre-control is cancelled, and the closed-loop control based on an air-fuel ratio is adopted; when the air-fuel ratio data is abnormal, the pre-control and the closed-loop control are directly cancelled, and an open-loop control mode is adopted.

It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.

The above-mentioned serial numbers of the embodiments of the present invention are merely for description and do not represent the merits of the embodiments.

While the present invention has been described with reference to the embodiments shown in the drawings, the present invention is not limited to the embodiments, which are illustrative and not restrictive, and it will be apparent to those skilled in the art that various changes and modifications can be made therein without departing from the spirit and scope of the invention as defined in the appended claims.