Machine learning based gain scheduling egr control method and electronic device

Abstract

The application discloses a gain scheduling EGR control method and electronic equipment based on machine learning, comprising: constructing an EGR gain scheduling control model; designing an EGR gain scheduling control model form; subdividing working condition points, solving the parameters of the EGR control model corresponding to each working condition point; dividing different working condition regions, and selecting the model with the highest fitting degree in each working condition region as the EGR control model of the current working condition region; constructing a gain scheduling coefficient MAP table; determining the gain scheduling coefficient MAP form; acquiring the gain scheduling MAP in the mode of machine learning; and dynamically adjusting the EGR control parameters based on real-time data. The application models the EGR control model for typical working condition points, then divides the working conditions of the engine operation into regions, selects the working condition point model with the highest fitting degree in each region, and then acquires the optimal control parameters by using the machine learning algorithm, and dynamically adjusts the EGR control parameters according to the real-time data, so that the control strategy of the engine is more accurate and adapts to different working conditions and driving requirements.

Description

Technical Field

[0001] This invention relates to the field of engine intake system control technology, specifically to a gain scheduling EGR control method and electronic device based on machine learning. Background Technology

[0002] In recent years, with the rapid growth of automobile production and ownership, the environmental pollution caused by motor vehicles has received widespread attention. Increasingly stringent emission regulations and fuel consumption standards have prompted automobile manufacturers to seek advanced engine auxiliary units and electronic control technologies to achieve energy conservation and emission reduction. EGR (Exhaust Gas Recirculation) technology, as a technology widely adopted by automobile manufacturers in recent years, has also seen its control issues become a research hotspot.

[0003] Exhaust gas recirculation (EGR) technology falls under the control of the engine's intake system, and its technical challenge lies in controlling the EGR rate. EGR technology reintroduces a portion of exhaust gas through the EGR valve back into the cylinder for combustion, reducing in-cylinder temperature and oxygen concentration, thus decreasing NOx emissions. Traditional EGR rate control methods generally employ MAP calibration combined with PID control for closed-loop control. PID control has the advantages of simple algorithms and ease of engineering implementation. Traditional EGR control strategies often rely on pre-set mapping tables or empirical rules, making them unable to adapt to changes in engine operating conditions.

[0004] The engine EGR system is a typical system with large hysteresis. Traditional EGR control usually uses PID control. Although PID control can achieve a relatively ideal control effect under the steady-state condition of the engine, it is prone to problems such as overshoot and slow adjustment process under the transient condition of the engine (a combination of a series of acceleration, deceleration, loading, unloading and idling conditions). Summary of the Invention

[0005] In view of the shortcomings of the existing technology, the purpose of this invention is to provide a gain scheduling EGR control method and electronic device based on machine learning, which aims to improve the accuracy of EGR rate control.

[0006] To solve the above-mentioned technical problems, the present invention is achieved through the following technical solution:

[0007] A gain scheduling (EGR) control method based on machine learning includes the following steps:

[0008] 1) Constructing the EGR gain scheduling control model:

[0009] 1.1) Design the EGR gain scheduling control model;

[0010] 1.2) Subdivide the operating points and solve for the parameters of the EGR control model corresponding to each operating point;

[0011] 1.3) Divide the operating conditions into different regions and select the model with the highest fitting degree in each operating condition region as the EGR control model for the current operating condition region;

[0012] 2) Construct a gain scheduling coefficient MAP table:

[0013] 2.1) Determine the MAP form of the gain scheduling coefficients;

[0014] 2.2) A gain scheduling MAP is obtained using machine learning.

[0015] 2.3) Based on the real-time data in steps 1.3) and 2.2), dynamically adjust the EGR control parameters.

[0016] Preferably, in step 1.1), the design of the EGR gain scheduling control model includes a transfer function and a state-space expression.

[0017] Preferably, in step 1.2), the step of solving the parameters of the EGR control model corresponding to each operating point is as follows:

[0018] Data from the engine at different operating conditions is collected, and machine learning is used to learn the parameters of the EGR control model that closely matches the actual EGR output value for each operating condition.

[0019] Preferably, data on the engine under different operating conditions are collected, including engine speed, torque, coolant temperature, EGR opening, exhaust bypass valve opening, and throttle opening.

[0020] Preferably, in step 1.3), the division of different operating condition zones is based on engine torque and speed.

[0021] Preferably, in step 2.1), the determination of the gain scheduling coefficient MAP is in the form of: selecting the engine speed, intake manifold pressure, and amplitude-frequency gain, and constructing a gain scheduling coefficient MAP table.

[0022] Preferably, in step 2.2), the step of obtaining the gain scheduling MAP using machine learning is as follows:

[0023] The optimal MAP value for each operating point is obtained through machine learning, and the optimal MAP values ​​for each operating point together form the gain scheduling MAP.

[0024] Preferably, in step 2.3), the EGR control parameters include the EGR opening degree, where EGR opening degree = EGR control model output × gain scheduling MAP.

[0025] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the method described above.

[0026] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described above.

[0027] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0028] 1. Model the EGR control model for typical operating conditions, then divide the engine operating conditions into regions, and select the operating condition model with the highest fitting degree for each region. The model has high accuracy and is easy to implement in engineering.

[0029] 2. By utilizing machine learning algorithms, the optimal control parameters are obtained, and the EGR control parameters are dynamically adjusted based on real-time data, making the engine control strategy more precise and adaptable to different operating conditions and driving requirements. Attached Figure Description

[0030] Figure 1 A detailed diagram for identifying operating points;

[0031] Figure 2 Bar charts showing the fitting degree of EGR valve models under different operating conditions;

[0032] Figure 3 A map showing the division of different working conditions and zones;

[0033] Figure 4 This is a MAP diagram for gain scheduling;

[0034] Figure 5 This is a framework diagram of the gain scheduling (EGR) control method based on machine learning.

[0035] Figure 6 The figure shows the simulation results of the gain scheduling (EGR) control method based on machine learning. Detailed Implementation

[0036] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0037] This invention provides a machine learning-based gain scheduling (EGR) control method and electronic device to improve the accuracy of EGR rate control.

[0038] To achieve the above-mentioned technical effects, the overall concept of this invention is as follows:

[0039] The key idea of ​​this invention is to use machine learning algorithms to model and predict the engine's operating state, and then adjust the gain scheduling of EGR control based on the prediction results to obtain the optimal EGR opening. Specifically, this invention includes the following steps:

[0040] I. Construction of EGR gain scheduling control model:

[0041] 1.1) Design the EGR gain scheduling control model, including but not limited to transfer functions, state-space expressions, etc.

[0042] 1.2) Control Model Parameter Solving: Data on engine performance at different operating points is collected, including engine speed, torque, coolant temperature, EGR opening, wastegate valve opening, and throttle opening. Machine learning methods (such as neural networks and support vector machines) are used to learn parameters that closely match the actual EGR output value to the EGR control model. When these parameters are input into the control model, under the same EGR opening signal input, the EGR control model outputs an EGR valve opening that closely approximates the actual value. Each operating point corresponds to a set of EGR valve control model parameters.

[0043] 1.3) Different operating conditions are divided into regions based on engine torque and speed. Within each operating condition region, the point with the highest fitting degree of the EGR valve model is selected as the typical operating condition point of that region. The control model parameters of that operating condition point are used, and the EGR control model of that operating condition point represents the EGR control model of that region.

[0044] II. Construction of Gain Scheduling Coefficient Map Table

[0045] 2.1) Determine the form of the gain scheduling coefficient MAP

[0046] The phase frequency characteristics of the EGR controller model at typical operating points in different regions are basically the same, but the amplitude frequency characteristics are different. Therefore, it is necessary to establish an amplitude frequency gain scheduling table and select the speed, intake manifold pressure, and amplitude frequency gain to construct a gain scheduling coefficient map table.

[0047] 2.2) Gain scheduling table acquisition

[0048] The gain scheduling table is formulated using machine learning methods (such as neural networks, reinforcement learning, etc.). The goal of machine learning is to minimize the error between the actual EGR rate and the target EGR rate by adjusting the EGR opening. The EGR opening is determined by the following formula: at the current operating point, the EGR opening equals the output of the EGR control model corresponding to the current operating point multiplied by the gain scheduling MAP. Through machine learning, the optimal MAP value corresponding to this operating point is obtained, which minimizes the error between the actual EGR rate and the target EGR rate.

[0049] EGR opening degree = EGR control model output × gain scheduling MAP

[0050] When the engine is running, the output value of the gain scheduling MAP changes with the current operating point to obtain the optimal EGR opening under the current operating condition.

[0051] The following is in conjunction with the appendix Figure 1-5 The embodiments of the present invention will be described in further detail below.

[0052] Step 1: Design the EGR gain scheduling control model

[0053] This example uses a transfer function to establish the system's physical model, with the following structure, where the number of poles is set to 2 and the number of zeros to 1:

[0054]

[0055] Y(s) represents the system output, U(s) represents the system input, and num(s) and den(s) are the numerator and denominator polynomials of the transfer function, respectively.

[0056] Step 2: Subdivide the operating points and solve for the control model parameters for each operating point.

[0057] like Figure 1 As shown, this example only subdivides 16 points as identification operating points, collects engine data under different operating points, and uses machine learning methods (such as neural networks and support vector machines) to fit the parameters of the EGR control model for each operating point, namely a0, a1, a2, b0, b1, ... Figure 2 This represents the degree of fit of the model parameters for each group.

[0058] Step 3: Division of Working Condition Zones and Selection of Typical Working Condition Models

[0059] like Figure 3 As shown, the operating condition region is divided into several regions based on speed and torque. Sixteen operating points are scattered across different regions. The model with the highest degree of fit is selected from each region as the EGR control model for the current operating condition region.

[0060] Step 4: Determine the gain scheduling MAP form

[0061] In this example, the amplitude-frequency gain scheduling table is constructed using engine speed, intake manifold pressure, and amplitude-frequency gain, and its form is as follows: Figure 4 As shown.

[0062] Step 5: Gain Scheduling MAP Acquisition

[0063] The gain scheduling MAP is obtained through machine learning. The gain scheduling value for each operating point obtained through machine learning is the optimal value. When this value is applied to the control system, the error between the actual EGR rate and the target EGR rate is minimized. The optimal gain scheduling values ​​for each operating point collectively constitute the gain scheduling MAP. The framework of the machine learning-based gain scheduling EGR control method is as follows: Figure 5 As shown.

[0064] Simulation has been completed, such as Figure 6 As shown, simulation results indicate that under the CLTC-P driving condition for light-duty vehicles in China, the air system controller based on Gain-Scheduling can meet the target of vehicle speed following, and the target vehicle speed is basically consistent with the actual vehicle speed; the actual EGR rate can follow the target EGR rate well, and the mean absolute error rate (MAPE) is less than 10%.

[0065] Based on the same inventive concept as the machine learning-based gain scheduling EGR control method embodiment, this embodiment of the invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the machine learning-based gain scheduling EGR control method as described in the above technical solution.

[0066] Based on the same inventive concept as the machine learning-based gain scheduling EGR control method embodiment, this embodiment of the invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the machine learning-based gain scheduling EGR control method as described in the above technical solution.

[0067] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the methods described above. Furthermore, any references to memory, databases, or other media used in the embodiments provided by this invention can include at least one of non-volatile and volatile memory.

[0068] The above are merely specific embodiments of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. A gain scheduling (EGR) control method based on machine learning, characterized in that, Includes the following steps: 1) Constructing the EGR gain scheduling control model: 1.1) Design the EGR gain scheduling control model; 1.2) Subdivide the operating points and solve the parameters of the EGR control model corresponding to each operating point; the steps for solving the parameters of the EGR control model corresponding to each operating point are as follows: collect engine data at different operating points, and learn the parameters of the EGR control model corresponding to each operating point that have a high degree of fit with the actual EGR output value through machine learning. 1.3) Divide the operating conditions into different regions based on engine torque and speed, and select the model with the highest fitting degree in each operating condition region as the EGR control model for the current operating condition region; 2) Construct a gain scheduling coefficient MAP table: 2.1) Determine the gain scheduling coefficient MAP form; the determination of the gain scheduling coefficient MAP form is as follows: select the engine speed, intake manifold pressure, and amplitude-frequency gain, and construct a gain scheduling coefficient MAP table; 2.2) A gain scheduling MAP is obtained using machine learning. 2.3) Based on the real-time data in steps 1.3) and 2.2), dynamically adjust the EGR control parameters.

2. The gain scheduling (EGR) control method based on machine learning according to claim 1, characterized in that: In step 1.1), the design of the EGR gain scheduling control model includes a transfer function and a state-space expression.

3. The gain scheduling (EGR) control method based on machine learning according to claim 1, characterized in that: Collect engine data at different operating conditions, including engine speed, torque, coolant temperature, EGR opening, exhaust bypass valve opening, and throttle opening.

4. The gain scheduling (EGR) control method based on machine learning according to claim 1, characterized in that: In step 2.2), the step of obtaining the gain scheduling MAP using machine learning is as follows: The optimal MAP value for each operating point is obtained through machine learning, and the optimal MAP values ​​for each operating point together form the gain scheduling MAP.

5. The gain scheduling (EGR) control method based on machine learning according to claim 1, characterized in that: In step 2.3), the EGR control parameters include the EGR opening degree, where EGR opening degree = EGR control model output × gain scheduling MAP.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, it implements the method as described in any one of claims 1-5.

7. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, it implements the method as described in any one of claims 1-5.

Images