阅读说明：本技术 一种基于插值计算的电喷系统闭环自学习控制方法 (Closed-loop self-learning control method of electronic fuel injection system based on interpolation calculation ) 是由 田李臣 蒋平 胡显力 曾军 张晓龙 谭聪 于 2020-06-04 设计创作，主要内容包括：一种基于插值计算的电喷系统闭环自学习控制方法,其是ECU基于发动机转速和负荷分为至少15个自学习区域,每个自学习区域使用一个自学习值,按照转速和负荷取区域的中间点为该自学习区域的区域基准点,将该区域自学习值赋予该区域基准点,对每个区域自学习更新需要提前判断合理性,合理后再写入,并且,自学习值读取时采用两次插值取值。采用本发明所述的控制方法,可以有效解决自学习偏离及邻近自学习值不同带来的空燃比跳变导致排放恶化问题,保证排放控制一致性。(An electronic fuel injection system closed-loop self-learning control method based on interpolation calculation is characterized in that an ECU is divided into at least 15 self-learning areas based on engine rotating speed and load, each self-learning area uses a self-learning value, the middle point of each self-learning area is taken as an area reference point of the self-learning area according to the rotating speed and the load, the area self-learning value is given to the area reference point, the reasonability needs to be judged in advance for self-learning updating of each area, writing is carried out after reasonability is achieved, and in addition, two times of interpolation value taking are adopted during self-learning value reading. By adopting the control method, the problem of emission deterioration caused by self-learning deviation and air-fuel ratio jump caused by different adjacent self-learning values can be effectively solved, and the consistency of emission control is ensured.)

1. The electronic fuel injection system closed-loop self-learning control method based on interpolation calculation is characterized by comprising the following steps: the control method is that the ECU is divided into at least 15 self-learning areas based on the rotating speed and the load of the engine, each self-learning area uses a self-learning value, the middle point of the area is taken as the area reference point of the self-learning area according to the rotating speed and the load, and the area self-learning value is endowed to the area reference point;

(1) a learning stage: after the engine is warmed up, the ECU calculates the deviation of the fuel injection quantity under the current working condition, namely a short-term self-learning value, based on the fuel injection control of the self-learning value of the datum point of the current region according to the feedback of the signal of the oxygen sensor; when the absolute value | short-term self-learning value-1 | of the change of the short-term self-learning value is larger than the self-learning control deviation for a certain time, the long-term self-learning value BLM of the current working condition point is calculated through the self-learning value interpolation of the reference points of the surrounding area_int(x，y)Comparing the short-term self-learning value of the current operating point with the long-term self-learning value BLM of the operating point_int(x，y)If the short-term self-learning value is positive, i.e. shortTerm self-learning value>1, and the short-term self-learning value is greater than the long-term self-learning value BLM_int(x，y)If the updating step length of one self-learning value is more than CLS, increasing the CLS for the self-learning value of the reference point of the area where the working point is located, and reducing the CLS for the short-term self-learning value; similarly, if the short-term self-learning value of the operating point is negative, the short-term self-learning value is<1, and the short-term self-learning value is smaller than the long-term self-learning value BLM of the working condition point_int(x，y)If the self-learning value updating step length is more than CLS, reducing the self-learning value of the reference point of the area where the working point is located by CLS, and increasing the short-term self-learning value by CLS;

(2) a reading stage: checking the nearest regional reference point condition of the current working point according to the engine speed load working point, and if 4 regional reference points are arranged around the current working point, performing secondary interpolation by using the self-learning values of the four regional reference points as the self-learning value of the current working point; if only two area reference points exist, performing one-time interpolation by using the self-learning values of the two area reference points as the self-learning value of the current working condition point; if only one area reference point exists, the self-learning value of the area reference point is used as the self-learning value of the current operating point.

2. An electronic fuel injection system closed-loop self-learning control method based on interpolation calculation as claimed in claim 1, wherein: the self-learning control deviation and the updating step length CLS are calibration values.

3. An electronic fuel injection system closed-loop self-learning control method based on interpolation calculation as claimed in claim 1, wherein: the time is determined by the number T of self-learning updating task scheduling time slices, and the T is a calibration value.

Technical Field

The invention belongs to the control technology of an automobile Engine Management System (EMS), and particularly relates to control of a Delftir electronic injection system.

Background

At present, national emission regulations are more and more strict, and national six is officially implemented in 2019, so that the emission stability is strengthened and delayed. Effective control of the engine electronic fuel injection system therefore directly affects batch vehicle emission levels. The Delfu electronic injection system (one of large suppliers of domestic electronic injection systems) adopts a self-learning control strategy, the self-learning of the Delfu electronic injection system is divided into short-term self-learning and long-term self-learning, the long-term self-learning adopts partition control, each area shares one self-learning value, the long-term self-learning value of the current operating working condition point directly uses the self-learning value of the area where the working condition point is located, the self-learning values of the areas are not particularly related, when the difference of the long-term self-learning values of two adjacent different areas is larger, the operating condition of an engine has self-learning value jump when the engine transits from one area to the other area, particularly when the working condition point transits between the two adjacent areas for multiple times, although certain filtering is set when the long-term self-learning value jumps, when the difference of the long-term self-learning values of the two adjacent different areas is larger, the air-, affecting emission levels.

Specifically, the existing delfu system self-learns by adopting partition control, and the running working condition of the engine is divided into fifteen areas according to different rotating speeds and loads to respectively carry out self-learning control. The batch vehicle emission level inspection shows that self-learning values of partial self-learning areas of partial vehicles are deformed due to vehicle conditions, oil products, service life, part degradation, data and the like, the self-learning values of two adjacent areas are different greatly, and the air-fuel ratio of the vehicles jumps in the variable working condition process to cause emission deterioration.

Disclosure of Invention

Aiming at the defects in the prior art, the electronic fuel injection system closed-loop self-learning control method based on interpolation calculation redefines the control strategy of the self-learning boundary region and redefines the data reading scheme, effectively solves the problem of emission deterioration caused by air-fuel ratio jump caused by self-learning deviation and different adjacent self-learning values, and ensures the consistency of emission control.

The technical scheme of the invention is as follows:

an electronic fuel injection system closed-loop self-learning control method based on interpolation calculation is characterized in that an ECU is divided into at least 15 self-learning areas based on rotating speed and load, the division is carried out according to control precision and ECU resource conditions, and if ECU resources are enough, the number of the control areas can be further expanded. And each self-learning area uses a self-learning value, the middle point of the area is taken as an area reference point of the self-learning area according to the rotating speed and the load, and the area self-learning value is endowed to the area reference point. The method comprises two parts:

(1) a learning stage: after the engine is warmed up, the ECU calculates the deviation of the fuel injection quantity under the current working condition based on the fuel injection control considering the self-learning value of the datum point of the current area according to the feedback of the signal of the oxygen sensor, namely the short-term self-learning value, when the absolute value | short-term self-learning value-1 | of the change of the short-term self-learning value is larger than the self-learning control deviation, and after a certain time, the long-term self-learning value BLM of the current working condition point is calculated_int(x，y)，BLM_int(x，y)And interpolating the self-learning values of the reference points of the surrounding area. Comparing the short-term self-learning value of the current operating point with the long-term self-learning value BLM of the operating point_int(x，y)If the short-term self-learning value is positive, i.e. short-term self-learning value>1, and the short term self-learning value is greater than the long term self-learning value (BLM)_int(x，y)If the updating step length of one self-learning value is more than CLS, increasing the CLS for the self-learning value of the reference point of the area where the working point is located, and reducing the CLS for the short-term self-learning value; similarly, if the short-term self-learning value of the operating point is negative, the short-term self-learning value is<1, and the short-term self-learning value is smaller than the long-term self-learning value BLM of the working condition point_int(x，y)And if the self-learning value updating step length is more than CLS, reducing the self-learning value of the reference point of the area where the working condition point is located by CLS, and increasing the short-term self-learning value by CLS.

(2) A reading stage: checking the nearest regional reference point condition of the current working point according to the engine speed load working point, and if 4 regional reference points (as shown in figure 2) are arranged around the current working point, performing secondary interpolation by using the self-learning values of the four regional reference points as the self-learning value of the current working point; if only two area reference points (such as fig. 3 and fig. 4) are available, performing one-time interpolation by using the self-learning values of the two area reference points to serve as the self-learning value of the current operating point; if only one area reference point exists (as shown in FIG. 5), the self-learning value of the area reference point is used as the self-learning value of the current operating point.

In the method, the self-learning control deviation and the updating step length CLS are calibration values. The time is determined by the number T of self-learning updating task scheduling time slices, and the T is a calibration value.

According to the control method, on the basis of the rotating speed load partition, the reasonability of each area needs to be judged in advance for self-learning updating, writing is carried out after reasonability is achieved, and two times of interpolation values are adopted during reading of self-learning values. Theoretically, the value of the read operation can be regarded as that all the partitions are fitted into a virtual continuous virtual table through two times of interpolation, namely, 2 times of interpolation is adopted in the area surrounded by the area reference points on the edge, one-time interpolation or direct value taking is adopted outside the area surrounded by the area reference points, and real-time updating is carried out, so that the control uniformity is ensured. Therefore, the control method can effectively solve the problem of emission deterioration caused by self-learning deviation and air-fuel ratio jump caused by different adjacent self-learning values, and ensure the consistency of emission control.

Drawings

FIG. 1 is a schematic illustration of a region reference point according to the present invention;

FIG. 2 is a schematic diagram of 4 region reference points around the current operating point (S);

FIG. 3 is a schematic diagram of the situation that 2 area reference points are around the current operating point (S) and belong to the same load area;

FIG. 4 is a schematic diagram of the situation that 2 reference points of the region are around the current operating point (S) and belong to the same rotating speed region;

fig. 5 is a schematic diagram of the case where there are 1 area reference points around the current operating point (S).

Detailed Description

The control method is further explained by combining the drawings and the specific embodiment as follows:

the closed-loop self-learning control method of the electronic injection system specifically comprises two parts of writing and reading of self-learning values.

As shown in FIG. 1, assuming that all operating points of the engine are divided into 16 areas, each area uses a self-learning value, the position of the self-learning value is assigned to each area as the rotating speed load position corresponding to the black point in FIG. 1, the self-learning value stored in the area where the current operating point is located is BLM (n), n is the number of the area where the current operating point is located, the current short-term self-learning value is CLI, the self-learning control deviation (the value can be calibrated, such as 0.03), and the updating step length of the self-learning value at the area reference point is CLS, wherein CLS is<Updating the number T of task scheduling time slices by the self-learning value (the time is determined by the number T of the self-learning updated task scheduling time slices, and T can be calibrated), and interpolating the self-learning value of the reference points of the surrounding area at the current working point to obtain the long-term self-learning value BLM_int(x，y)。

The self-learning update procedure uses the following formula:

the self-learning value reading is divided into three cases: checking the nearest reference point condition of the surrounding area according to the rotating speed load working point

1. Four regional reference points are around:

as shown in fig. 2, 4 area reference points are provided around the operating point S, which are area reference points 1, 2, 5, and 6, and when the engine is operating at the operating point S, the system performs secondary interpolation based on the rotation speed and the load using the self-learning values BLM (1), BLM (2), BLM (5), and BLM (6) corresponding to the area reference points 1, 2, 5, and 6 as the self-learning values of the current operating point.

2. There are two regional reference points around:

1) as shown in FIG. 3, 2 area reference points are arranged around the operating point S, namely area reference points 1 and 5, and when the engine runs at the operating point S, the system uses the self-learning values BLM (1) and BLM (5) corresponding to the area reference points 1 and 5 to perform primary interpolation based on the rotating speed to serve as the self-learning values of the current operating point.

2) As shown in FIG. 4, 2 area reference points are arranged around the operating point S, namely area reference points 13 and 14, and when the engine runs at the operating point S, the system uses the self-learning values BLM (13) and BLM (14) corresponding to the area reference points 13 and 14 as the self-learning values of the current operating point based on the primary interpolation of the load.

3. There are 1 regional fiducial around:

as shown in fig. 5, there are only 1 area reference point around the operating point S, i.e., the area reference point 13, and when the engine is operating at the operating point S, the system directly uses the self-learning value BLM (13) corresponding to the area reference point 13 without interpolation.

The control method can be realized by software, and needs to be implemented in a writing code embedded system based on a corresponding electronic injection system.