Automobile engine dynamic control method and device based on neural network and FPGA

By introducing neural networks and FPGAs into the engine control system, the engine operating status prediction model is constructed, which solves the problem of insufficient flexibility of traditional systems in complex environments, realizes intelligent and efficient engine control, and improves driving safety and environment Protection effect.

CN119021795AActive Publication Date: 2024-11-26SOUTHWEST UNIV

Patent Information

Application Number
CN202411034809.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-31
Publication Date
2024-11-26
Estimated Expiration
2044-07-31

AI Technical Summary

Technical Problem

When facing a complex and changing driving environment, traditional engine control systems are difficult to flexibly deal with emergencies and unknown variables. The sensor data processing method is simple, and it cannot fully utilize the potential correlations and deep-seated features in the data, affecting driving safety and Comfort.

Method used

The engine dynamic control method based on neural network and FPGA is adopted. By designing the wiring scheme corresponding to the operating states of different engines, an engine operating state prediction model is built, and the model is trained using historical data to predict and switch the wiring scheme in real time to control the operating state of the engine.

Benefits of technology

It improves the intelligence level and response speed of engine control, achieves flexible adaptation to complex driving environments, improves the efficiency and stability of the control system, reduces energy waste and emissions, and improves driving comfort and safety.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119021795A_ABST
    Figure CN119021795A_ABST
Patent Text Reader

Abstract

The invention discloses an automobile engine dynamic control method and device based on a neural network and an FPGA, and the method comprises the following steps: designing a wiring scheme of the FPGA, and enabling different engine operation states to correspond to different wiring schemes; arranging a wiring selector, and constructing an engine running state prediction model in the wiring selector; obtaining historical data of vehicle operation, and training the engine operation state prediction model by using the historical data of vehicle operation; using the trained engine operation state prediction model to predict the engine operation state in real time; and controlling the FPGA to switch to a corresponding wiring scheme according to the engine operation state predicted by the engine operation state prediction model so as to control the engine to reach the predicted engine operation state. According to the invention, the neural network model is introduced and the FPGA is combined on the basis, so that the method and the device which can deal with a complicated and changeable driving environment and intelligently and efficiently control the running state of the engine can be provided.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of automobile technology, and in particular to a dynamic control of an automobile engine based on a neural network and FPGA. Background Art

[0002] With the rapid development of the automobile industry, the driving environment faced by vehicles is becoming increasingly complex and changeable, which poses unprecedented challenges to the intelligence, precision and rapid response capabilities of engine control systems. Although traditional engine control systems have achieved remarkable results in improving fuel efficiency and reducing emissions, their core still relies on preset control logic (not decision-making through neural network models) and fixed sensor data processing modes. This mode performs well when dealing with static or predictable environmental changes, but its flexibility and adaptability are insufficient when faced with complex factors such as rapidly changing traffic conditions, diverse driving habits and extreme weather conditions on actual roads. Traditional control systems have the following defects: Traditional control systems often design a set of fixed control logic based on engineers' experience and experimental data, which is difficult to flexibly respond to emergencies and unknown variables in actual driving; the processing method of sensor data is relatively simple, and often only limited information can be extracted for control decisions, ignoring the potential correlation and deep-level features in the data; when faced with scenarios that require rapid decision-making, traditional systems may be unable to make timely adjustments due to the limitations of computational complexity, affecting driving safety and comfort.

[0003] Therefore, there is an urgent need to provide an engine dynamic control method that can break through the limitations of traditional control systems. Summary of the invention

[0004] In view of the above-mentioned deficiencies in the prior art, the technical problem to be solved by the present invention is to provide a vehicle engine dynamic control method and device based on neural network and FPGA, which can cope with complex and changeable driving environments and control the engine operation more intelligently and efficiently.

[0005] In order to solve the above technical problems, a technical solution adopted by the present invention is: to provide a vehicle engine dynamic control method based on neural network and FPGA, comprising the following steps:

[0006] Design the wiring scheme of FPGA. Different wiring schemes correspond to different engine operation states.

[0007] Setting a wiring selector, and constructing an engine operation state prediction model in the wiring selector;

[0008] Acquiring historical data of vehicle operation, and using the historical data of vehicle operation to train the engine operation state prediction model;

[0009] Use the trained engine operation status prediction model to predict the engine operation status in real time;

[0010] According to the engine operating state predicted by the engine operating state prediction model, the FPGA is controlled to switch to a corresponding wiring scheme to control the engine to achieve the predicted engine operating state.

[0011] Furthermore, the engine operating state prediction model includes an input layer, three hidden layers and an output layer;

[0012] The input layer is used to input a feature vector;

[0013] The first hidden layer is used to process the input feature vector and output a first set of nonlinear features by calculating the weighted sum and the bias term;

[0014] The second hidden layer is used to perform a nonlinear transformation on the output of the first hidden layer, add a Dropout layer to prevent overfitting, and output a second set of nonlinear features;

[0015] The third hidden layer is used to ensure the generalization ability and nonlinear processing ability of the model, and outputs a third set of nonlinear features as the basis for the final prediction;

[0016] The output layer is used to obtain the probability distribution of each engine operating state according to the third group of nonlinear features and output the engine operating state with the greatest probability.

[0017] Furthermore, the step of acquiring the historical data of vehicle operation and using the historical data of vehicle operation to train the engine operation state prediction model specifically includes the following sub-steps:

[0018] Acquire historical data of vehicle operation to form a database, wherein the historical data includes sensor data, vehicle operation parameters and corresponding engine operation status;

[0019] Preprocessing the data in the database to obtain sample data;

[0020] The sample data is used to train an engine operating state prediction model.

[0021] Furthermore, the step of obtaining historical data and forming a database specifically includes the following sub-steps:

[0022] Defining labels for different engine operating states, the labels including starting state, idling state, accelerating state, light load state, medium load state, and heavy load state;

[0023] Drive the vehicle under different road conditions to collect sensor data and vehicle operating parameters;

[0024] For each data sample collected, it is accurately associated with a predefined label according to the actual operating state of the engine at the time of collection, and each sample is labeled to form a database.

[0025] Furthermore, the sensor data includes camera data, lidar data, ultrasonic sensor data and inertial measurement unit data; the vehicle operating parameters include vehicle speed, steering angle and battery power;

[0026] The step of preprocessing the data in the database to obtain sample data specifically includes the following sub-steps:

[0027] Clean the data in the database and remove abnormal data;

[0028] Process and extract features of camera data, lidar data, ultrasonic sensor data, and inertial measurement unit data respectively;

[0029] The features extracted from the camera data, lidar data, ultrasonic sensor data, and inertial measurement unit data are combined with the vehicle's speed, steering angle, and battery power to obtain a feature vector.

[0030] Assign a corresponding label of the engine operating status to each feature vector to obtain sample data.

[0031] Furthermore, the method for processing the camera data and extracting features is: firstly performing image preprocessing on the photos taken by the camera and then extracting image features from the preprocessed images, wherein the image features include edges, textures and shapes;

[0032] The method for processing laser radar data and extracting features is: firstly reducing the noise of the point cloud image constructed by the laser radar through median filtering or mean filtering, and then extracting features from the denoised point cloud image, wherein the features include density and curvature;

[0033] The method for processing ultrasonic sensor data and extracting features is as follows: using a sliding window to smooth the ultrasonic signal, and then determining the actual distance based on the time difference between the transmitted wave and the echo;

[0034] The method for processing the inertial measurement unit data and extracting features is to use a sliding window to smooth the acceleration data and integrate the smoothed acceleration data to obtain velocity and displacement.

[0035] Furthermore, the step of performing image preprocessing on the photos taken by the camera specifically includes the following sub-steps:

[0036] Grayscale processing: converting color images into grayscale images to reduce the amount of calculation and model complexity;

[0037] Resizing, resizing the image to a fixed size to ensure the consistency of the input data;

[0038] Normalization processing: normalize the pixel values ​​to between 0 and 1 to improve the convergence speed and training effect of the model;

[0039] The step of extracting image features from the preprocessed image specifically includes: using a convolutional neural network to extract features from the preprocessed image.

[0040] Furthermore, the step of using the sample data to train the engine operating state prediction model specifically includes the following sub-steps:

[0041] Inputting training data including feature vectors and expected outputs into the engine operation state prediction model, wherein the training data is obtained by dividing the sample data;

[0042] Extracting features related to the engine operating state according to the feature vector, and converting the extracted features into probabilities of each engine operating state;

[0043] Selecting the engine operating state with the highest probability as the output of the output layer of the engine operating state prediction model;

[0044] The error between the output of the output layer and the expected output is calculated, and the error is used for back propagation to adjust the model parameters until the model converges.

[0045] Further, the step of controlling the FPGA to switch to a corresponding wiring scheme to control the engine to reach the predicted engine operating state according to the engine operating state predicted by the engine operating state prediction model specifically includes the following sub-steps:

[0046] Send the predicted engine operation status data to the state machine in the FPGA in real time;

[0047] The state machine in the FPGA selects a wiring scheme that matches the predicted engine operating state according to preset mapping rules;

[0048] After the state machine selects the wiring scheme, the FPGA's wiring resources are dynamically reconfigured through the logic control unit inside the FPGA to switch to the selected wiring scheme, so that the engine reaches the predicted operating state.

[0049] In order to solve the above technical problems, another technical solution adopted by the present invention is to provide a vehicle engine dynamic control device based on a neural network and FPGA, comprising:

[0050] Design wiring module, used to design FPGA wiring scheme, different engine operation states correspond to different wiring schemes;

[0051] A model building module, used for setting a wiring selector and building an engine operation state prediction model in the wiring selector;

[0052] A data acquisition and model training module, used to acquire historical data of vehicle operation and train the engine operation state prediction model using the historical data of vehicle operation;

[0053] A prediction module, used to predict the engine operating state in real time using a trained engine operating state prediction model;

[0054] The control module is used to control the FPGA to switch to a corresponding wiring scheme according to the engine operating state predicted by the engine operating state prediction model to control the engine to achieve the predicted engine operating state.

[0055] The automobile engine dynamic control method and device based on neural network and FPGA of the present invention have at least the following beneficial effects: by introducing a neural network model, the present invention can make full use of a large amount of historical data of vehicle operation, automatically learn and identify the optimal operating state of the engine under different driving conditions. This intelligent decision-making mechanism not only improves the accuracy and response speed of the system, but also enables the system to adapt to complex and changeable driving environments in real time, and can flexibly deal with various uncertain factors, thereby significantly improving the intelligent level of engine control; combined with the high parallel processing capability and real-time response characteristics of FPGA, the present invention can quickly execute the prediction results of the neural network model and realize the rapid dynamic switching of the engine control circuit. This hardware-level switching speed far exceeds that of traditional The software control method ensures seamless transition of the engine between different states, effectively improving the efficiency and stability of the control system. At the same time, precise control also reduces energy waste and emissions, which is of great significance to environmental protection and energy conservation and emission reduction. The programmable characteristics of FPGA make the engine control system of the present invention extremely flexible and scalable. As the driving environment and requirements continue to change, the control logic can be quickly optimized and adjusted by reconfiguring the logic circuit inside the FPGA to adapt to new operating conditions and requirements. By controlling the engine operating state in real time and intelligently, the driving comfort and safety can be significantly improved. In different driving scenarios, the engine can quickly respond to the driver's intentions and the actual needs of the vehicle to provide stable and efficient power output. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0057] Figure 1 The flowchart is a method for dynamic control of an automobile engine based on a neural network and FPGA according to an embodiment of the present invention.

[0058] Figure 2 for Figure 1 Flow chart of step S3 in FIG.

[0059] Figure 3 for Figure 2 Flowchart of step S31 in FIG.

[0060] Figure 4 for Figure 2 Flowchart of step S32 in FIG.

[0061] Figure 5 for Figure 2 Flowchart of step S33 in FIG.

[0062] Figure 6 for Figure 1 Flow chart of step S5 in FIG.

[0063] Figure 7 The flowchart of one embodiment of the automobile engine dynamic control device based on neural network and FPGA of the present invention. DETAILED DESCRIPTION

[0064] The present invention will be further described below in conjunction with the accompanying drawings.

[0065] See also Figure 1 , is a flow chart of an embodiment of a method for dynamic control of an automobile engine based on a neural network and FPGA of the present invention. This embodiment specifically includes the following steps:

[0066] S1. Design the wiring scheme of FPGA. Different engine operating states correspond to different wiring schemes.

[0067] In the present embodiment, the engine operating state includes a starting state, an idle state, an acceleration state, a low load state, a medium load state, and a high load state. It should be understood that for different vehicle models, the parameter ranges of each engine operating state are different. In the present embodiment, the starting state refers to a state where the engine speed is greater than or equal to 1200 rpm and less than or equal to 1500 rpm, the speed is 0, and the throttle opening is greater than or equal to 5% and less than or equal to 8%; the idle state refers to a state where the engine speed is greater than or equal to 600 rpm and less than or equal to 1000 rpm, the speed is 0, and the throttle opening is greater than or equal to 3% and less than or equal to 5%; the acceleration state refers to a state where the engine speed is greater than or equal to 3000 rpm and the throttle opening is greater than or equal to 60% and less than or equal to 100%; the low load state refers to a state where the engine speed is greater than or equal to 1200 rpm and the throttle opening is greater than or equal to 1500 rpm; The medium load state refers to the state when the engine speed is greater than or equal to 2000 rpm and less than or equal to 3500 rpm, the speed is greater than or equal to 50 km / h and less than or equal to 80 km / h, and the throttle opening is greater than or equal to 20% and less than or equal to 40%; the heavy load state refers to the state when the engine speed is greater than or equal to 3500 rpm and less than or equal to 4500 rpm, the speed is greater than or equal to 80 km / h and less than or equal to 120 km / h, and the throttle opening is greater than or equal to 40% and less than or equal to 60%.

[0068] S2. Setting a wiring selector and constructing an engine operating state prediction model in the wiring selector.

[0069] Specifically, the engine operation state prediction model includes an input layer, three hidden layers and an output layer. The input layer is used to input feature vectors; the first hidden layer is used to process the input feature vectors, and output a first group of nonlinear features by calculating the weighted sum and the bias term. In this embodiment, the first hidden layer includes 64 neurons, and the input feature vectors are processed using the ReLU activation function. The first group of nonlinear features includes 64 features. The second hidden layer is used to perform a nonlinear transformation on the output of the first hidden layer, add a Dropout layer to prevent overfitting, and output a second group of nonlinear features. In this embodiment, the second hidden layer includes 128 neurons, and the first group of nonlinear features output by the first hidden layer is processed using the ReLU activation function. A Dropout layer with a Dropout layer of 0.5 is added, and the second group of nonlinear features includes 128 features. The third hidden layer is used to ensure the generalization ability and nonlinear processing ability of the model, and outputs the third group of nonlinear features as the basis for the final prediction. In this embodiment, the third hidden layer includes 64 neurons, and uses the ReLU activation function to process the second group of nonlinear features output by the second hidden layer. A 0.5 Dropout layer is added to prevent overfitting. The third group of nonlinear features includes 64 features. The output layer is used to obtain the probability distribution of each engine operating state according to the third group of nonlinear features and output the engine operating state with the highest probability. In this embodiment, the output layer includes 6 neurons, and uses the Softmax activation function to convert the third group of nonlinear features into the probability distribution of 6 engine operating states and output the engine operating state with the highest probability.

[0070] S3. Acquire historical data of vehicle operation, and use the historical data of vehicle operation to train the engine operation state prediction model.

[0071] See also Figure 2 , this step S3 includes the following sub-steps:

[0072] S31. Acquire historical data of vehicle operation to form a database, wherein the historical data includes sensor data, vehicle operation parameters, and corresponding engine operation status.

[0073] Specifically, the sensor data includes camera data, laser radar data, ultrasonic sensor data and inertial measurement unit data. The camera data is detected by a camera sensor, and the camera data includes information about traffic signs, lane lines, pedestrians and other vehicles. The vehicle needs to adjust its speed and driving direction based on this information; the laser radar data is detected by a laser radar sensor, and the laser radar data includes high-precision environmental map information and obstacle information. According to the obstacle density in the surrounding environment, the vehicle can decide whether to slow down or avoid; the ultrasonic sensor data is detected by an ultrasonic sensor, and the ultrasonic sensor is used to detect distance, such as distance control when parking. According to the parking scene, the vehicle can decide whether to slow down or stop; the inertial measurement unit data is detected by an inertial measurement unit sensor, and the inertial measurement unit data includes the acceleration, angular velocity and direction of the vehicle. According to the dynamic changes of the vehicle, the vehicle can adjust the acceleration and deceleration.

[0074] The vehicle operating parameters include the vehicle's speed, steering angle, and battery power. Based on the vehicle's current speed and target speed, the vehicle can decide whether to accelerate or decelerate; based on the vehicle's steering angle, the vehicle can adjust the speed and turning radius; based on the current battery power, the vehicle can decide whether to accelerate or decelerate. For example, when the vehicle's battery power is low, the vehicle needs to decelerate to extend the range.

[0075] See also Figure 3 , this step S31 includes the following sub-steps:

[0076] S311. Define labels for different engine operating states, wherein the labels include a start state, an idle state, an acceleration state, a light load state, a medium load state, and a heavy load state.

[0077] S312. Drive the vehicle under different road conditions and collect sensor data and vehicle operating parameters.

[0078] Specifically, the road conditions include but are not limited to urban roads, highways, mountainous areas, rainy days and nights, and data collection is performed under various road scenarios to ensure that the data set can comprehensively cover various driving environments and engine operating conditions.

[0079] S313. For each collected data sample, accurately associate it with a predefined label according to the actual operating state of the engine when it is collected, complete the labeling of each sample, and form a database.

[0080] In order to solve the problem of unbalanced data sets and improve the generalization ability and accuracy of subsequent model training, in this implementation, oversampling (copying existing samples or generating new samples), undersampling (reducing the number of samples of a larger category), synthetic data generation or other technical means are also used to ensure that the number of samples corresponding to different labels is roughly equal. However, it should be noted that oversampling may lead to overfitting, while undersampling may lose important information. Therefore, different weights can also be assigned to samples of different categories when training the model to compensate for the impact of data imbalance.

[0081] S32: Preprocess the data in the database to obtain sample data.

[0082] See also Figure 4 , this step S32 includes the following sub-steps:

[0083] S321. Clean the data in the database and remove abnormal data.

[0084] Different cleaning strategies are used for different data problems. Specifically, duplicate data is deleted, and duplicate data in the database is checked and deleted; missing values ​​are processed, and missing values ​​in the data are identified, and statistical methods such as average, maximum, and minimum are used to fill in the data or delete records containing missing values; outlier detection and processing, outliers are detected using statistical methods or business rules, and then corrected or deleted.

[0085] S322, process the camera data, the lidar data, the ultrasonic sensor data, and the inertial measurement unit data respectively and extract features.

[0086] Specifically, the method for processing the camera data and extracting features is: first preprocessing the photos taken by the camera and then extracting image features from the preprocessed images, the image features include edges, textures and shapes. In this embodiment, a convolutional neural network is used to extract features from the preprocessed images. Among them, the method for image preprocessing the photos taken by the camera is: grayscale processing, converting color images into grayscale images to reduce the amount of calculation and model complexity; resizing, adjusting the image to a fixed size to ensure the consistency of the input data; normalization processing, normalizing the pixel values ​​to between 0 and 1, improving the convergence speed and training effect of the model.

[0087] The method for processing and extracting features of laser radar data is: first reduce the noise of the point cloud map constructed by the laser radar through median filtering or mean filtering, and then extract features from the denoised point cloud map, wherein the features include density and curvature. The method for processing and extracting features of ultrasonic sensor data is: use a sliding window to smooth the ultrasonic signal, and then determine the actual distance based on the time difference between the transmitted wave and the echo. The method for processing and extracting features of inertial measurement unit data is: use a sliding window to smooth the acceleration data, and integrate the smoothed acceleration data to obtain velocity and displacement.

[0088] S323, perform feature fusion to obtain a feature vector.

[0089] Specifically, the features extracted from the camera data, lidar data, ultrasonic sensor data, and inertial measurement unit data are fused with the vehicle's speed, steering angle, and battery power to obtain a feature vector.

[0090] S324: assign a corresponding label of the engine operating status to each feature vector to obtain sample data.

[0091] S33: Using the sample data to train an engine operating state prediction model.

[0092] See also Figure 5 , this step S33 includes the following sub-steps:

[0093] S331, inputting training data including feature vectors and expected outputs into the engine operating state prediction model, wherein the training data is obtained by dividing sample data.

[0094] Specifically, the feature vector includes features extracted from camera data, lidar data, ultrasonic sensor data, and inertial measurement unit data, as well as the vehicle's speed, steering angle, and battery power. The desired output is the engine operating state as a label. The sample data is divided into training data and test data in a certain ratio. The training data is used to train the engine operating state prediction model, and the test data is used to evaluate the performance of the trained model on unknown data. By dividing the training data and the test data, it can be ensured that the model will not overfit the training data during the training process, that is, the model will not only remember the features of the training data but ignore the inherent laws of the data.

[0095] S332. Extract features related to the engine operating state according to the feature vector, and convert the extracted features into probabilities of various engine operating states.

[0096] Specifically, according to the input vector, the output of the hidden layer and the output layer is calculated, that is, the forward propagation process. The input vector is processed by the ReLU activation function to generate the output of the hidden layer. The output of the hidden layer is then passed to the output layer and processed by the Softmax activation function to obtain the probability distribution of different engine operating states.

[0097] S333. Select the engine operating state with the highest probability as the output of the output layer of the engine operating state prediction model.

[0098] S334. Calculate the error between the output of the output layer and the expected output, and use the error to perform back propagation and adjust the model parameters until the model converges.

[0099] S4. Use the trained engine operating state prediction model to predict the engine operating state in real time.

[0100] Specifically, sensor data and vehicle operating parameters are acquired in real time, and the sensor data and vehicle operating parameters are preprocessed to obtain feature vectors, which are input into the input layer of the engine operating state prediction model. After being processed by three hidden layers, the third group of nonlinear features is passed to the output layer as the basis for the final prediction. The output layer converts the third group of nonlinear features into the probability distribution of 6 engine operating states and outputs the engine operating state with the highest probability as the prediction result.

[0101] S5. According to the engine operating state predicted by the engine operating state prediction model, the FPGA is controlled to switch to a corresponding wiring scheme to control the engine to reach the predicted engine operating state.

[0102] See also Figure 6 , this step S5 includes the following sub-steps:

[0103] S51, sending the predicted engine operating status data to the state machine in the FPGA in real time.

[0104] Specifically, the wiring selector receives sensor data and vehicle operating parameters, uses a neural network model to predict the engine operating status, and outputs a data signal corresponding to the engine operating status, which is transmitted to the state machine in the FPGA. The signal is a binary signal. For example, if the output is the sixth state, the signal sent to the state machine is 110.

[0105] S52. The state machine in the FPGA selects a wiring scheme that matches the predicted engine operating state according to a preset mapping rule.

[0106] Specifically, after the state machine in the FPGA receives the binary signal output by the wiring selector, it parses the binary signal, identifies the predicted engine operating state, and compares and analyzes it according to the preset mapping rules to find the corresponding wiring scheme. The mapping rules predefine the correspondence between each engine operating state and a specific wiring scheme to ensure a quick and accurate response to changes in the engine state.

[0107] S53: Switch to a wiring scheme that matches the predicted engine operating state.

[0108] Specifically, after the state machine selects a wiring scheme, the FPGA's wiring resources are dynamically reconfigured through the logic control unit inside the FPGA to switch to the selected wiring scheme, so that the engine reaches the predicted operating state. Switching the wiring scheme can change the internal logic circuit of the FPGA to implement different control algorithms and logic functions, thereby affecting the actions of various actuators in the engine control system and the operating state of the engine. For example, by adjusting the parameters or logic flow of the control algorithm in the FPGA, it is possible to achieve precise control of key parameters such as injection quantity and ignition advance angle, thereby changing the operating state of the engine. Switching the wiring scheme can also change the connection relationship of the external devices of the FPGA, thereby affecting the synergy of the external devices and achieving control of the engine's operating state.

[0109] See also Figure 7 , is a flow chart of an embodiment of a vehicle engine dynamic control device based on a neural network and FPGA of the present invention. The vehicle engine dynamic control device based on a neural network and FPGA of this embodiment is used to implement the vehicle engine dynamic control method based on a neural network and FPGA as described in the above embodiment. Specifically, the vehicle engine dynamic control device based on a neural network and FPGA of this embodiment includes a design wiring module 100, a model building module 200, a data acquisition and model training module 300, a prediction module 400 and a control module 500. Among them:

[0110] The design wiring module 100 is used to design the wiring scheme of the FPGA, and different engine operating states correspond to different wiring schemes. In this embodiment, the engine operating state includes a start state, an idle state, an acceleration state, a small load state, a medium load state, and a large load state.

[0111] The model building module 200 is used to set a wiring selector and build an engine operating state prediction model in the wiring selector.

[0112] The data acquisition and model training module 300 is used to acquire the historical data of vehicle operation, and use the historical data of vehicle operation to train the engine operation state prediction model. Specifically, the data acquisition and model training module 300 includes a data acquisition module 310, a preprocessing module 320 and a model training module 330, wherein: the data acquisition module 310 is used to acquire the historical data of vehicle operation to form a database, and the historical data includes sensor data, vehicle operation parameters and corresponding engine operation status; the preprocessing module 320 is used to preprocess the data in the database formed by the data acquisition module 310 to obtain sample data; the model training module 330 is used to train the engine operation state prediction model using the sample data.

[0113] The prediction module 400 is used to predict the engine operating state in real time using a trained engine operating state prediction model.

[0114] The control module 500 is used to control the FPGA to switch to a corresponding wiring scheme according to the engine operating state predicted by the engine operating state prediction model to control the engine to reach the predicted engine operating state.

[0115] By introducing a neural network model, the present invention can make full use of a large amount of historical data on vehicle operation, automatically learn and identify the optimal operating state of the engine under different driving conditions. This intelligent decision-making mechanism not only improves the accuracy and response speed of the system, but also enables the system to adapt to complex and changeable driving environments in real time and flexibly respond to various uncertain factors, thereby significantly improving the intelligence level of engine control. Combined with the high parallel processing capability and real-time response characteristics of FPGA, the present invention can quickly execute the prediction results of the neural network model and realize fast dynamic switching of the engine control circuit. This hardware-level switching speed far exceeds the traditional software control method, ensuring the seamless transition of the engine between different states. The engine control method of the present invention has high flexibility and scalability due to the programmable characteristics of FPGA. With the continuous changes in driving environment and demand, the control logic can be quickly optimized and adjusted by reconfiguring the logic circuit inside FPGA to adapt to new operating conditions and demand. The real-time and intelligent control of the engine running state can significantly improve the driving comfort and safety. In different driving scenarios, the engine can quickly respond to the driver's intentions and the actual needs of the vehicle to provide stable and efficient power output.

Claims

1. A method for dynamic control of automobile engine based on neural network and FPGA, characterized in that: The following steps are involved: Design the wiring scheme of FPGA. Different wiring schemes correspond to different engine operation states. Setting a wiring selector, and constructing an engine operation state prediction model in the wiring selector; Acquiring historical data of vehicle operation, and using the historical data of vehicle operation to train the engine operation state prediction model; Use the trained engine operation status prediction model to predict the engine operation status in real time; According to the engine operating state predicted by the engine operating state prediction model, the FPGA is controlled to switch to a corresponding wiring scheme to control the engine to achieve the predicted engine operating state.

2. The automobile engine dynamic control method based on neural network and FPGA as claimed in claim 1, characterized in that: The engine operating state prediction model includes an input layer, three hidden layers and an output layer; The input layer is used to input a feature vector; The first hidden layer is used to process the input feature vector and output a first set of nonlinear features by calculating the weighted sum and the bias term; The second hidden layer is used to perform a nonlinear transformation on the output of the first hidden layer, add a Dropout layer to prevent overfitting, and output a second set of nonlinear features; The third hidden layer is used to ensure the generalization ability and nonlinear processing ability of the model, and outputs a third set of nonlinear features as the basis for the final prediction; The output layer is used to obtain the probability distribution of each engine operating state according to the third group of nonlinear features and output the engine operating state with the greatest probability.

3. The automobile engine dynamic control method based on neural network and FPGA as claimed in claim 1, characterized in that: The step of acquiring the historical data of vehicle operation and using the historical data of vehicle operation to train the engine operation state prediction model specifically includes the following sub-steps: Acquire historical data of vehicle operation to form a database, wherein the historical data includes sensor data, vehicle operation parameters and corresponding engine operation status; Preprocessing the data in the database to obtain sample data; The sample data is used to train an engine operating state prediction model.

4. The automobile engine dynamic control method based on neural network and FPGA as claimed in claim 3, wherein the step of acquiring historical data and forming a database specifically comprises the following sub-steps: Defining labels for different engine operating states, the labels including starting state, idling state, accelerating state, light load state, medium load state, and heavy load state; Drive the vehicle under different road conditions to collect sensor data and vehicle operating parameters; For each data sample collected, it is accurately associated with a predefined label according to the actual operating state of the engine at the time of collection, and each sample is labeled to form a database.

5. The automobile engine dynamic control method based on neural network and FPGA as claimed in claim 4, characterized in that: The sensor data includes camera data, lidar data, ultrasonic sensor data and inertial measurement unit data; the vehicle operation parameters include vehicle speed, steering angle and battery power; The step of preprocessing the data in the database to obtain sample data specifically includes the following sub-steps: Clean the data in the database and remove abnormal data; Process and extract features of camera data, lidar data, ultrasonic sensor data, and inertial measurement unit data respectively; The features extracted from the camera data, lidar data, ultrasonic sensor data, and inertial measurement unit data are combined with the vehicle's speed, steering angle, and battery power to obtain a feature vector. Assign a corresponding label of the engine operating status to each feature vector to obtain sample data.

6. The automobile engine dynamic control method based on neural network and FPGA as claimed in claim 5, characterized in that: The method for processing the camera data and extracting features is as follows: firstly performing image preprocessing on the photos taken by the camera and then extracting image features from the preprocessed images, wherein the image features include edges, textures and shapes; The method for processing laser radar data and extracting features is: firstly reducing the noise of the point cloud image constructed by the laser radar through median filtering or mean filtering, and then extracting features from the denoised point cloud image, wherein the features include density and curvature; The method for processing ultrasonic sensor data and extracting features is as follows: using a sliding window to smooth the ultrasonic signal, and then determining the actual distance based on the time difference between the transmitted wave and the echo; The method for processing the inertial measurement unit data and extracting features is to use a sliding window to smooth the acceleration data and integrate the smoothed acceleration data to obtain velocity and displacement.

7. The automobile engine dynamic control method based on neural network and FPGA as claimed in claim 6, characterized in that: The step of performing image preprocessing on the photos taken by the camera specifically includes the following sub-steps: Grayscale processing: converting color images into grayscale images to reduce the amount of calculation and model complexity; Resizing, resizing the image to a fixed size to ensure the consistency of the input data; Normalization processing: normalize the pixel values ​​to between 0 and 1 to improve the convergence speed and training effect of the model; The step of extracting image features from the preprocessed image specifically includes: using a convolutional neural network to extract features from the preprocessed image.

8. The automobile engine dynamic control method based on neural network and FPGA as claimed in claim 5, characterized in that: The step of using the sample data to train the engine operating state prediction model specifically includes the following sub-steps: Inputting training data including feature vectors and expected outputs into the engine operation state prediction model, wherein the training data is obtained by dividing the sample data; Extracting features related to the engine operating state according to the feature vector, and converting the extracted features into probabilities of each engine operating state; Selecting the engine operating state with the highest probability as the output of the output layer of the engine operating state prediction model; The error between the output of the output layer and the expected output is calculated, and the error is used for back propagation to adjust the model parameters until the model converges.

9. The automobile engine dynamic control method based on neural network and FPGA as claimed in claim 1, characterized in that: The step of controlling the FPGA to switch to a corresponding wiring scheme according to the engine operating state predicted by the engine operating state prediction model to control the engine to reach the predicted engine operating state specifically includes the following sub-steps: Send the predicted engine operation status data to the state machine in the FPGA in real time; The state machine in the FPGA selects a wiring scheme that matches the predicted engine operating state according to preset mapping rules; After the state machine selects the wiring scheme, the FPGA's wiring resources are dynamically reconfigured through the logic control unit inside the FPGA to switch to the selected wiring scheme, so that the engine reaches the predicted operating state.

10. A vehicle engine dynamic control device based on neural network and FPGA, characterized in that: include: Design wiring module, used to design FPGA wiring scheme, different engine operation states correspond to different wiring schemes; A model building module, used for setting a wiring selector and building an engine operation state prediction model in the wiring selector; A data acquisition and model training module, used to acquire historical data of vehicle operation and train the engine operation state prediction model using the historical data of vehicle operation; A prediction module, used to predict the engine operating state in real time using a trained engine operating state prediction model; The control module is used to control the FPGA to switch to a corresponding wiring scheme according to the engine operating state predicted by the engine operating state prediction model to control the engine to achieve the predicted engine operating state.

Citation Information

Patent Citations

  • Control device and method of internal combustion engine, in-vehicle electronic control unit and manufacturing method of electronic control unit, machine learning system, and output parameter calculation device

    CN111520241A

  • Aero-engine thrust calculation method, storage medium and calculation equipment

    CN113128047A

  • Method and device for optimally parameterizing driving dynamics control system of vehicle

    CN115534970A

  • Engine running state prediction method and system and electronic equipment

    CN117390510A

  • Control method and controller for predicting aircraft model based on neural network

    CN117555270A

Cited By

  • Vehicle running state real-time detection method and system

    CN120183063A

  • Control method for self-adaptive optimization of injection parameters and improvement of combustion efficiency of fuel system

    CN121952752A