Vehicle control method, vehicle, electronic device and readable medium
By setting up a compensation device in the advanced assisted driving system, using neural network models to process the required torque and send error compensation torque, the time delay and accuracy error problems of electric power steering system are solved, and the accuracy and reliability of vehicle driving control are improved.
Patent Information
- Application Number
- CN202411061304.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-05
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2044-08-05
AI Technical Summary
The existing electric power steering system has time lag and execution accuracy errors, resulting in reduced accuracy and reliability of vehicle driving control.
A compensation device is set up in an advanced assisted driving system, and the required torque is processed using the target neural network model, error compensation torque information is obtained, and sent to the electronic power steering system to improve the accuracy and reliability of the actual torque.
Through the input of error compensation torque information, the actual torque output by the electronic power steering system is close to the required torque, which improves the output accuracy and reliability of the system, and does not require modification of the existing electronic power steering system, making it easy to implement engineering.
Smart Images

Figure CN118560571B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of vehicle technology, and in particular to a vehicle control method, a vehicle, an electronic device and a computer-readable medium. Background Art
[0002] In the prior art, the Electric Power Steering (EPS) system usually has time lag and execution accuracy error. Among them, time lag usually refers to the delay time between the driver starting to turn the steering wheel and the vehicle actually starting to turn. The execution accuracy error usually refers to the torque actually output by the electric power steering system may have a certain error with the torque required by the vehicle. Time lag and execution accuracy error may reduce the accuracy and reliability of vehicle driving control. Summary of the invention
[0003] The embodiment of the present invention provides a vehicle control method, a vehicle, an electronic device and a computer-readable storage medium to improve the accuracy and reliability of vehicle driving control.
[0004] An embodiment of the present invention discloses a vehicle control method, which is applied to a compensation device, wherein the compensation device is arranged in an advanced driver assistance system of a vehicle, and the vehicle further includes an electronic power steering system. The method includes:
[0005] Obtaining required torque information from the advanced driver assistance system;
[0006] Inputting the required torque information into a target neural network model preset in the compensation device, and obtaining the error compensation torque information output by the target neural network model;
[0007] The error compensation torque information is sent to the electronic power steering system; the electronic power steering system is used to output an actual torque consistent with the required torque information based on the error compensation torque information.
[0008] Optionally, the method further comprises:
[0009] Get vehicle speed information;
[0010] The step of inputting the required torque information into a target neural network model preset in the compensation device and obtaining the error compensation torque information output by the target neural network model comprises:
[0011] The required torque information and the vehicle speed information are input into a target neural network model preset in the compensation device, and the error compensation torque information output by the target neural network model is obtained.
[0012] Optionally, the method further comprises:
[0013] Acquire a preset training data set, wherein the training data set includes at least one training sample; the training sample includes input sample information and a standard output torque; the input sample information includes a required torque sample;
[0014] The training samples are used to train a preset neural network model to obtain a trained target neural network model.
[0015] Optionally, the input sample information also includes a vehicle speed sample.
[0016] Optionally, the demand torque samples include steady-state torque samples and / or dynamic torque samples.
[0017] Optionally, the neural network model is provided with model parameters;
[0018] The step of using the training samples to train a preset neural network model to obtain a trained target neural network model includes:
[0019] Inputting the input sample information into a neural network model to obtain a model output torque output by the neural network model;
[0020] Adjusting the model parameters based on the difference between the model output torque and the standard output torque corresponding to the input sample information;
[0021] Repeat the steps of inputting the input sample information into the neural network model, obtaining the model output torque output by the neural network model, and adjusting the model parameters based on the difference between the model output torque and the standard output torque corresponding to the input sample information, until the number of repeated executions reaches a preset training number threshold, or the difference between the model output torque and the standard output torque corresponding to the input sample information is less than a preset error threshold, thereby obtaining a trained target neural network model.
[0022] Optionally, the neural network includes an input layer, a hidden layer connected to the input layer, and an output layer connected to the hidden layer;
[0023] The model parameters include a hidden layer weight matrix and a hidden layer bias matrix corresponding to the hidden layer, and also include an input layer weight matrix and an input layer bias matrix corresponding to the output layer;
[0024] The step of inputting the input sample information into a neural network model to obtain a model output torque output by the neural network model comprises:
[0025] Inputting the input sample information into the input layer respectively to obtain the first data output by the input layer;
[0026] Inputting the first data into the hidden layer to obtain second data calculated by the hidden layer based on the first data, the hidden layer weight matrix and the hidden layer bias matrix;
[0027] The second data is input into the output layer to obtain the model output torque calculated by the output layer based on the second data, the output layer weight matrix and the output layer bias matrix.
[0028] Optionally, the input layer includes m input layer nodes, and the hidden layer includes n hidden layer nodes; wherein m≥1, n≥1; the first data includes outputs of the m input layer nodes; the hidden layer weight matrix includes weights corresponding to the hidden layer nodes, and the hidden layer bias matrix includes biases corresponding to the hidden layer nodes;
[0029] The step of inputting the first data into the hidden layer to obtain second data calculated by the hidden layer based on the first data, the hidden layer weight matrix and the hidden layer bias matrix comprises:
[0030] For the nth hidden layer node, respectively obtain the outputs of the m input layer nodes;
[0031] Using a hyperbolic tangent function, a weight corresponding to the nth hidden layer node, and a bias corresponding to the nth hidden layer node, the outputs of the m input layer nodes are processed respectively to obtain m first processed data;
[0032] The sum of the m first processed data is used as the output of the nth hidden layer node, and the outputs of the n hidden layer nodes are used as the second data.
[0033] Optionally, the step of inputting the second data into the output layer to obtain a model output torque calculated by the output layer based on the second data, the output layer weight matrix, and the output layer bias matrix includes:
[0034] For the output layer, respectively obtain the outputs of the n hidden layer nodes;
[0035] Using a linear function, the output layer weight matrix, and the output layer bias matrix, the outputs of the n hidden layer nodes are processed respectively to obtain n second processed data;
[0036] The sum of the n second processed data is used as the model output torque output by the output layer.
[0037] The embodiment of the present invention further discloses a vehicle control device, which is applied to a compensation device, wherein the compensation device is arranged in an advanced driver assistance system of a vehicle, wherein the vehicle further includes an electronic power steering system, and the device includes:
[0038] A demand torque acquisition module, used to obtain demand torque information from the advanced driver assistance system;
[0039] An error compensation torque determination module, used for inputting the required torque information into a target neural network model preset in the compensation device, and obtaining the error compensation torque information output by the target neural network model;
[0040] The torque output module is used to send the error compensation torque information to the electronic power steering system; the electronic power steering system is used to output an actual torque consistent with the required torque information based on the error compensation torque information.
[0041] Optionally, the device further comprises:
[0042] A vehicle speed acquisition module is used to obtain vehicle speed information;
[0043] The error compensation torque determination module comprises:
[0044] The error compensation torque determination submodule is used to input the required torque information and the vehicle speed information into a target neural network model preset in the compensation device, and obtain the error compensation torque information output by the target neural network model.
[0045] Optionally, the device further comprises:
[0046] A training data acquisition module, used to acquire a preset training data set, wherein the training data set includes at least one training sample; the training sample includes input sample information and a standard output torque; the input sample information includes a required torque sample;
[0047] The model training module is used to use the training samples to train the preset neural network model to obtain a trained target neural network model.
[0048] Optionally, the input sample information also includes a vehicle speed sample.
[0049] Optionally, the demand torque samples include steady-state torque samples and / or dynamic torque samples.
[0050] Optionally, the neural network model is provided with model parameters;
[0051] The model training module includes:
[0052] A model output acquisition submodule, used to input the input sample information into the neural network model to obtain the model output torque output by the neural network model;
[0053] A parameter adjustment submodule, used for adjusting the model parameters based on the difference between the model output torque and the standard output torque corresponding to the input sample information;
[0054] A repeated training submodule is used to repeatedly execute the steps of inputting the input sample information into the neural network model, obtaining the model output torque output by the neural network model, and adjusting the model parameters based on the difference between the model output torque and the standard output torque corresponding to the input sample information, until the number of repeated executions reaches a preset training number threshold, or the difference between the model output torque and the standard output torque corresponding to the input sample information is less than a preset error threshold, thereby obtaining a trained target neural network model.
[0055] Optionally, the neural network includes an input layer, a hidden layer connected to the input layer, and an output layer connected to the hidden layer;
[0056] The model parameters include a hidden layer weight matrix and a hidden layer bias matrix corresponding to the hidden layer, and also include an input layer weight matrix and an input layer bias matrix corresponding to the output layer;
[0057] The model output acquisition submodule includes:
[0058] A first data acquisition unit, used for inputting the input sample information into the input layer respectively to obtain first data output by the input layer;
[0059] A second data acquisition unit, configured to input the first data into the hidden layer, and obtain second data calculated by the hidden layer based on the first data, the hidden layer weight matrix, and the hidden layer bias matrix;
[0060] A model output torque acquisition unit is used to input the second data into the output layer to obtain the model output torque calculated by the output layer based on the second data, the output layer weight matrix and the output layer bias matrix.
[0061] Optionally, the input layer includes m input layer nodes, and the hidden layer includes n hidden layer nodes; wherein m≥1, n≥1; the first data includes outputs of the m input layer nodes; the hidden layer weight matrix includes weights corresponding to the hidden layer nodes, and the hidden layer bias matrix includes biases corresponding to the hidden layer nodes;
[0062] The second data acquisition unit includes:
[0063] A first node output acquisition subunit, used for acquiring the outputs of the m input layer nodes respectively for the nth hidden layer node;
[0064] A first data processing subunit is used to process the outputs of the m input layer nodes respectively by using a hyperbolic tangent function, a weight corresponding to the nth hidden layer node, and a bias corresponding to the nth hidden layer node to obtain m first processed data;
[0065] The second data determination subunit is used to use the sum of the m first processed data as the output of the nth hidden layer node, and use the outputs of the n hidden layer nodes as the second data.
[0066] Optionally, the model output torque acquisition unit includes:
[0067] A second node output acquisition subunit, used for respectively acquiring outputs of the n hidden layer nodes for the output layer;
[0068] A second data processing subunit is used to process the outputs of the n hidden layer nodes respectively using a linear function, the output layer weight matrix, and the output layer bias matrix to obtain n second processed data;
[0069] The model output torque determination subunit is used to take the sum of the n second processed data as the model output torque output by the output layer.
[0070] The embodiment of the present invention further discloses a vehicle, comprising a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus;
[0071] The memory is used to store computer programs;
[0072] The processor is used to implement the method described in any one of the embodiments of the present invention when executing the program stored in the memory.
[0073] The embodiment of the present invention further discloses an electronic device, comprising a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus;
[0074] The memory is used to store computer programs;
[0075] The processor is used to implement the method described in the embodiment of the present invention when executing the program stored in the memory.
[0076] The embodiment of the present invention further discloses one or more computer-readable media on which instructions are stored. When executed by one or more processors, the processors are enabled to execute the method described in the embodiment of the present invention.
[0077] The embodiments of the present invention include the following advantages:
[0078] Through the vehicle control method provided by the embodiment of the present invention, the required torque information is obtained from the advanced driver assistance system; the required torque information is input into the target neural network model preset in the compensation device, and the error compensation torque information output by the target neural network model is obtained; the error compensation torque information is sent to the electronic power steering system; the electronic power steering system is used to output an actual torque consistent with the required torque information based on the error compensation torque information. By inputting torque information with error compensation into the electronic power steering system, the actual torque output by the electronic power steering system based on the error compensation torque information can be close to the original required torque, thereby improving the accuracy and reliability of the output of the electronic power steering system. At the same time, the embodiment of the present invention only needs to add a compensation device to the advanced driver assistance system, without modifying the existing electronic power steering system, which is easy to implement in engineering. BRIEF DESCRIPTION OF THE DRAWINGS
[0079] Figure 1 It is a schematic diagram of the required torque and the actual torque output by the electronic power steering system in the prior art;
[0080] Figure 2 is a flow chart of steps of a vehicle control method provided in an embodiment of the present invention;
[0081] Figure 3 It is a schematic diagram of the relationship between the required torque and the actual output torque of the electronic power steering system EPS in the prior art;
[0082] Figure 4 It is a schematic diagram of the relationship between vehicle speed and actual output torque of the electronic power steering system EPS in the prior art;
[0083] Figure 5 is a flow chart of a vehicle control method provided by an embodiment of the present invention;
[0084] Figure 6 is a comparison test diagram of the required torque and the actual torque output by the electronic power steering system in the embodiment of the present invention;
[0085] Figure 7 is a comparison test diagram of the required torque and the actual torque output by the electronic power steering system in the embodiment of the present invention;
[0086] Figure 8is a structural block diagram of a vehicle control device provided in an embodiment of the present invention;
[0087] Fig. 9 It is a block diagram of an electronic device provided in an embodiment of the present invention. DETAILED DESCRIPTION
[0088] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.
[0089] In the prior art, the required torque of the vehicle and the torque actually output by the electronic power steering system can be expressed as follows: Figure 1 As shown in Figure 2, it can be seen that there is always a certain error between the required torque of the vehicle and the torque actually output by the electronic power steering system. This error may be caused by time lag or by the accuracy error of the actual output of the electronic power steering system.
[0090] Specifically, the required torque required by the vehicle may include at least a steady-state required torque and a dynamic required torque. Specifically, in order to allow the vehicle to travel in a circle at the same speed relatively stably, the vehicle usually needs to periodically adjust the steering angle to compensate for changes in the steering torque caused by factors such as lateral force, tire characteristics, and suspension system, so that when performing stable steering, the vehicle can usually generate a steady-state required torque that is close to a sine wave. When the vehicle needs to perform emergency obstacle avoidance according to actual road conditions, or when auxiliary driving systems in the vehicle such as lane keeping assist systems, adaptive cruise control systems, and electronic body stability systems need to adjust the vehicle's driving trajectory, the vehicle can generate non-fixed torque requirements, thereby forming a dynamic required torque.
[0091] Specifically, Figure 1 (a) is a comparison diagram between the required torque input to the electronic power steering system EPS and the actual output torque of the electronic power steering system EPS when the vehicle generates a steady-state required torque. Figure 1 (c) shows Figure 1 The enlarged view of 25s-33s in (a). It can be seen that when the demand torque is a steady-state demand torque, if the amplitude of the steady-state demand torque is small, such as 25-29s, there may be a large precision error between the actual output torque and the steady-state demand torque, and there is also a time lag. When the amplitude of the steady-state demand torque is large, such as 29-33s, there may be a small precision error between the actual output torque and the steady-state demand torque, and there is mainly a time lag.
[0092] Figure 1(b) is a comparison diagram between the required torque input to the electronic power steering system EPS and the actual output torque of the electronic power steering system EPS when the vehicle generates a dynamic required torque. Figure 1 (d) shows Figure 1 The enlarged view of (b) is from 25s to 38s. It can be seen that when the required torque is a dynamic torque, there may be a large precision error between the actual output torque and the dynamic required torque, and there is also a time lag.
[0093] In summary, due to time lag and precision error, the required torque of the vehicle may differ from the torque actually output by the electronic power steering system, which reduces the accuracy and reliability of vehicle control. It is necessary to control the electronic power steering system to output a more accurate actual torque. In particular, the precision error causes a relatively large gap between the actual output torque and the required torque, so it is necessary to improve the accuracy of the actual output torque of the electronic power steering system.
[0094] In the case where the gap between the actual output torque and the required torque is relatively large, the embodiment of the present invention sets a compensation device in the advanced driver assistance system, processes the required torque through the target neural network model in the compensation device, obtains error compensation torque information, and inputs the error compensation torque information into the electronic power steering system. The actual torque output by the electronic power steering system based on the error compensation torque information can be close to the original required torque, thereby improving the accuracy and reliability of the output of the electronic power steering system by inputting torque information with error compensation into the electronic power steering system. At the same time, the embodiment of the present invention only needs to add a compensation device in the advanced driver assistance system, without modifying the existing electronic power steering system, and is easy to implement in engineering.
[0095] Reference Figure 2 , shows a flow chart of steps of a vehicle control method provided in an embodiment of the present invention, which is applied to a compensation device, the compensation device is arranged in an advanced driver assistance system of a vehicle, and the vehicle also includes an electronic power steering system. The method may specifically include the following steps:
[0096] Step 201, obtaining required torque information from an advanced driver assistance system;
[0097] In a specific implementation, the advanced driver assistance system can collect the required torque generated by the driver through controlling the steering wheel, or collect the current operating status of the vehicle and the surrounding environment of the vehicle based on sensors in the vehicle such as cameras, laser radars, ultrasonic sensors, wheel speed sensors, steering angle sensors, acceleration sensors, gyroscopes, etc., and determine the required torque required to keep the vehicle running stably based on the current operating status of the vehicle, the surrounding environment of the vehicle, and the vehicle control information generated by the driver. Afterwards, the required torque is sent to the electronic power steering system to control the electronic power steering system to apply the required torque to control the vehicle to run stably.
[0098] Specifically, the actual torque output by the electronic power steering system can be strongly correlated with the required torque information output by the advanced driver assistance system. Figure 3 As shown, Figure 3 This is a schematic diagram of the relationship between the required torque and the actual output torque of the electronic power steering system EPS in the prior art. It can be seen that in response to different required torques, the actual output torque of the electronic power steering system EPS can fluctuate within a certain range accordingly. For example, when the required torque is -2Nm, the actual output torque of the electronic power steering system EPS can fluctuate around -2Nm. When the required torque is 0Nm, the actual output torque of the electronic power steering system EPS can fluctuate within the range of -1Nm~1Nm.
[0099] Therefore, in order to improve the accuracy of the actual torque output by the electronic power steering system, the advanced driver assistance system can send the required torque to the compensation device before inputting the required torque into the electronic power steering system, and adjust the required torque through the compensation device, so that the subsequent actual output of the electronic power steering system can be more accurate.
[0100] Step 202, inputting the required torque information into a target neural network model preset in the compensation device, and obtaining the error compensation torque information output by the target neural network model;
[0101] In a specific implementation, a target neural network model may be provided in the compensation device. The target neural network model may be trained to take the required torque as input and output error compensation torque information. The error compensation torque information has error compensation based on the required torque, and can compensate for the error generated by the electronic power steering system. Thus, after the compensation device obtains the required torque information from the advanced driver assistance system, it inputs the obtained required torque information into the target neural network model, and obtains the error compensation torque information output by the target neural network model.
[0102] Step 203: Send the error compensation torque information to the electronic power steering system; the electronic power steering system is used to output an actual torque consistent with the required torque information based on the error compensation torque information.
[0103] Thereafter, the compensation device may send the error compensation torque information to the electronic power steering system. The electronic power steering system may output the actual torque based on the error compensation torque information. The actual torque may differ from the error compensation torque, but after the error compensation process, the actual torque output by the electronic power steering system may be close to the required torque originally determined by the advanced driver assistance system, so that the required torque and the actual torque may be as consistent as possible, thereby improving the accuracy and reliability of the output of the electronic power steering system.
[0104] Through the vehicle control method provided by the embodiment of the present invention, the required torque information is obtained from the advanced driver assistance system; the required torque information is input into the target neural network model preset in the compensation device, and the error compensation torque information output by the target neural network model is obtained; the error compensation torque information is sent to the electronic power steering system; the electronic power steering system is used to output the actual torque consistent with the required torque information based on the error compensation torque information. By inputting the torque information with error compensation into the electronic power steering system, the actual torque output by the electronic power steering system based on the error compensation torque information can be close to the original required torque, thereby improving the accuracy and reliability of the output of the electronic power steering system. At the same time, the embodiment of the present invention only needs to add a compensation device to the advanced driver assistance system, without modifying the existing electronic power steering system, and is easy to implement in engineering.
[0105] In one embodiment of the present invention, the method further comprises:
[0106] S11, obtaining vehicle speed information;
[0107] In specific implementations, when the electronic power steering system outputs torque, in addition to considering the required torque, it can also further consider the current speed of the vehicle. For example, when driving at low speeds, the vehicle usually needs a larger torque output to make steering easier. When driving at high speeds, in order to maintain the stability and controllability of the vehicle, the torque output can be relatively reduced to make the steering more stable. Therefore, different vehicle speeds can correspond to different torque control strategies.
[0108] like Figure 4 As shown, Figure 4It is a schematic diagram of the relationship between the vehicle speed and the actual output torque of the electronic power steering system EPS in the prior art. It can be seen that at different vehicle speeds, the range of the actual output torque of the electronic power steering system EPS may be different. For example, at a vehicle speed of 65km / h, the range of the actual output torque of the electronic power steering system EPS may be between -1.5Nm and 2Nm. At a vehicle speed of 75km / h, the range of the actual output torque of the electronic power steering system EPS may be between -2Nm and 2Nm. At a vehicle speed of 80km / h, the range of the actual output torque of the electronic power steering system EPS may be between -1Nm and 2Nm. It can be seen that for the electronic power steering system EPS, different vehicle speeds may correspond to different torque control strategies.
[0109] It can be seen that for the electronic power steering system, the factors that affect the actual torque output include not only the required torque but also the vehicle speed. There is a strong correlation between the vehicle speed and the actual torque output by the electronic power steering system. In order to more accurately compensate the error of the electronic power steering system and improve the accuracy of the actual torque output by the electronic power steering system, the compensation device can further consider the vehicle speed information in the process of determining the error compensation torque information. In this way, the vehicle speed information can be obtained.
[0110] In one embodiment of the present invention, the step of inputting the required torque information into a target neural network model preset in the compensation device and obtaining the error compensation torque information output by the target neural network model includes:
[0111] S21, inputting the required torque information and the vehicle speed information into a target neural network model preset in the compensation device, and obtaining the error compensation torque information output by the target neural network model.
[0112] In the compensation device, the required torque information and the vehicle speed information can be used as inputs of the target neural network model at the same time, and the error compensation torque information output by the target neural network model based on the required torque information and the vehicle speed information can be obtained. At the same time, the error compensation torque information output based on the required torque information and the vehicle speed information can have higher accuracy, and better improve the accuracy of the actual torque output by the electronic power steering system.
[0113] As a specific example of the present invention, Figure 5 It is a flow chart of a vehicle control method provided by an embodiment of the present invention.
[0114] After the advanced driver assistance system 501 calculates the required torque information based on the original control algorithm, the required torque information is input into the compensation device 502 .
[0115] The compensation device 502 can obtain vehicle speed information and input the required torque information and the vehicle speed information into the target neural network model. The target neural network model can output error compensation torque information based on the input required torque information and the vehicle speed information. The error compensation torque information is input into the electronic power steering system 503.
[0116] Based on the error compensation torque information and the vehicle speed information, the electronic power steering system 503 can determine the actual torque that is consistent with the required torque output by the advanced driver assistance system 501, and apply the actual torque to the vehicle, thereby achieving precise control of the vehicle torque.
[0117] In one embodiment of the present invention, the method further comprises:
[0118] S31, obtaining a preset training data set, wherein the training data set includes at least one training sample; the training sample includes input sample information and a standard output torque; the input sample information includes a required torque sample;
[0119] In a specific implementation, in order to enable the target neural network model to have the ability to output error compensation torque information based on the required torque information. A neural network model can be trained first. Thus, a training data set for training the neural network model can be obtained. The training data set may include at least one training sample. Each training sample includes input sample information as input, and as a standard output, a standard output torque corresponding to each input sample information. Among them, the input sample information may include a required torque sample.
[0120] Optionally, when it is necessary to further improve the accuracy of the model output, the input sample information may further include a vehicle speed sample, so that the trained model can take the required torque and the vehicle speed as input at the same time.
[0121] Optionally, since the torque generated by the vehicle under different working conditions has different states, in order to enable the neural network model to more comprehensively cover different working conditions, the required torque sample may specifically include a steady-state torque sample and / or a dynamic torque sample, so that the electronic power steering system can output the actual torque more accurately under the steady-state required torque and / or the dynamic required torque.
[0122] S32, using the training samples to train a preset neural network model to obtain a trained target neural network model.
[0123] Thereafter, the training samples in the training data set may be used to train the preset neural network model, so that the neural network model can output error compensation torque information based on the required torque information, thereby obtaining a trained target neural network model.
[0124] In a specific implementation, the neural network model can be a backpropagation neural network model (BP neural network), a recurrent neural network model (RNN), a long short-term memory network model (LSTM), etc., and the present invention does not limit this.
[0125] Optionally, the neural network model is provided with model parameters. Model parameters are variables that the neural network model needs to learn and adjust during training.
[0126] The step of using the training samples to train a preset neural network model to obtain a trained target neural network model includes:
[0127] S41, inputting the input sample information into a neural network model to obtain a model output torque output by the neural network model;
[0128] In a specific implementation, after the input sample information is input into the neural network model, the neural network model can output a model output torque based on the input sample information.
[0129] S42, adjusting the model parameters based on the difference between the model output torque and the standard output torque corresponding to the input sample information;
[0130] For the training samples in the training data set, the standard output torque corresponding to the input sample information can be recorded. The standard output torque can be considered as the correct output that the neural network model needs to learn. Therefore, the model output torque can be compared with the standard output torque to calculate the difference between the two. And the back propagation method is adopted to adjust the model parameters in the neural network model based on the difference between the model output torque and the standard output torque, so that the model output torque output by the neural network can be closer to the standard output torque.
[0131] The difference between the model output torque and the standard output torque may be represented by mean square error (MSE), mean absolute error (MAE), etc., and the present invention does not impose any limitation on this.
[0132] S43, repeatedly inputting the input sample information into the neural network model, obtaining the model output torque output by the neural network model, and adjusting the model parameters based on the difference between the model output torque and the standard output torque corresponding to the input sample information, until the number of repeated executions reaches a preset training number threshold, or the difference between the model output torque and the standard output torque corresponding to the input sample information is less than a preset error threshold, thereby obtaining a trained target neural network model.
[0133] Afterwards, the training process of the neural network model can be repeated, the input sample information can be input into the neural network model, the model output torque output by the neural network model can be obtained, and the model parameters can be adjusted based on the difference between the model output torque and the standard output torque corresponding to the input sample information, so that the model output torque of the neural network continues to approach the standard output torque.
[0134] When the number of repeated executions of the training process of the neural network model reaches the preset training number threshold, or the difference between the model output torque of the neural network model and the standard output torque corresponding to the input sample information is less than the preset error threshold, it can be considered that the neural network model has achieved a good training effect. At this time, the neural network model training can be considered completed, and a trained target neural network model is obtained.
[0135] Optionally, the learning rate of the neural network model training process can also be set. The learning rate represents the update step size of the model parameters during each retraining of the model. When the learning rate is large, the model parameters will be updated at a larger amplitude, which can make the model output torque output by the neural network model approach the standard output torque faster, but the optimal model parameters may be missed due to the large update amplitude. When the learning rate is small, the model parameters can be updated at a smaller amplitude, which may cause the neural network model to converge too slowly, increase the number of training times, and thus increase the training time, but it can make the model parameters closer to the optimal situation. Therefore, the appropriate learning rate can be set according to actual needs so that the trained neural network model has a better effect.
[0136] Optionally, the neural network includes an input layer, a hidden layer connected to the input layer, and an output layer connected to the hidden layer;
[0137] The model parameters include a hidden layer weight matrix and a hidden layer bias matrix corresponding to the hidden layer, and also include an input layer weight matrix and an input layer bias matrix corresponding to the output layer;
[0138] The step of inputting the input sample information into the neural network model to obtain the model output torque output by the neural network model includes:
[0139] S51, inputting the input sample information into the input layer respectively to obtain first data output by the input layer;
[0140] In the specific implementation, the input layer is mainly used to receive input data. After receiving the input sample information, the input sample information can be directly output. The input sample information can also be normalized and then output.
[0141] Specifically, when the input sample information includes demand torque samples and vehicle speed samples, the numerical range of the demand torque may be at a different order of magnitude from the numerical range of the vehicle speed. For example, the demand torque may generally be in the range of -3~3Nm, while the numerical range of the vehicle speed may be in the range of 0~200km / h. In this case, due to the different numerical magnitudes, if the input sample information is directly output, it is easy to cause the subsequent neural network model to ignore the demand torque with a smaller magnitude during data processing, resulting in poor model training results. Therefore, the demand torque samples and vehicle speed samples can be normalized, and the two can be mapped to the same numerical range respectively, so that the neural network model can simultaneously focus on the influence of the demand torque and vehicle speed on the actual torque during subsequent data processing.
[0142] S52, inputting the first data into the hidden layer to obtain second data calculated by the hidden layer based on the first data, the hidden layer weight matrix and the hidden layer bias matrix;
[0143] The hidden layer may have a hidden layer weight matrix or a hidden layer bias matrix, which are model parameters that can be adjusted during the training of the neural network model. The hidden layer may be calculated based on the first data, the hidden layer weight matrix or the hidden layer bias matrix to obtain second data as the hidden layer output.
[0144] S53, inputting the second data into the output layer to obtain the model output torque calculated by the output layer based on the second data, the output layer weight matrix and the output layer bias matrix.
[0145] The output layer may have an output layer weight matrix and an output layer bias matrix, which are model parameters that can be adjusted during the training process of the neural network model. The output layer may be calculated based on the second data, the output layer weight matrix and the output layer bias matrix to obtain a model output torque as the output of the output layer.
[0146] In one embodiment of the present invention, the input layer includes m input layer nodes, and the hidden layer includes n hidden layer nodes; wherein m≥1, n≥1; the first data includes outputs of the m input layer nodes; the hidden layer weight matrix includes weights corresponding to the hidden layer nodes, and the hidden layer bias matrix includes biases corresponding to the hidden layer nodes;
[0147] The step of inputting the first data into the hidden layer to obtain second data calculated by the hidden layer based on the first data, the hidden layer weight matrix and the hidden layer bias matrix comprises:
[0148] S61, for the nth hidden layer node, respectively obtain the outputs of the m input layer nodes;
[0149] S62, using a hyperbolic tangent function, a weight corresponding to the nth hidden layer node, and a bias corresponding to the nth hidden layer node, respectively processing the outputs of the m input layer nodes to obtain m first processed data;
[0150] For the nth hidden layer node, in order to calculate the output of the node, the outputs of the m input layer nodes can be obtained respectively. The output of the input layer node can be the input sample information or the input sample information after normalization.
[0151] The weight corresponding to the nth hidden layer node may include the weight corresponding to each input layer node. The bias corresponding to the nth hidden layer node may include the bias corresponding to each input layer node.
[0152] The hyperbolic tangent function can be the activation function of the hidden layer, which can be expressed as:
[0153]
[0154] For the output of the mth input layer node, it can be first multiplied by the weight corresponding to the input layer node, then added with a bias, and then calculated using a hyperbolic tangent function to obtain the first processed data.
[0155] S63, taking the sum of the m first processed data as the output of the nth hidden layer node, and taking the outputs of the n hidden layer nodes as the second data.
[0156] After respectively calculating and obtaining the first processed data corresponding to each input layer node, the first processed data may be added to obtain the output of the nth hidden layer node.
[0157] As a specific example of the present invention, when there are 7 input layer nodes, for the nth hidden layer node, the output of the node can be expressed as:
[0158]
[0159] in, is the output of the nth hidden layer node of the neural network, is the mth input of the hidden layer node of the neural network, which is calculated by multiplying the output of the mth input layer node by the weight and then adding the bias.
[0160] In one embodiment of the present invention, the step of inputting the second data into the output layer to obtain the model output torque calculated by the output layer based on the second data, the output layer weight matrix and the output layer bias matrix includes:
[0161] S71, for the output layer, respectively obtaining the outputs of the n hidden layer nodes;
[0162] S72, using a linear function, the output layer weight matrix, and the output layer bias matrix to process the outputs of the n hidden layer nodes respectively to obtain n second processed data;
[0163] For the output layer, it may have one node. In order to calculate the output of the node, the outputs of n hidden layer nodes may be obtained respectively.
[0164] The weight corresponding to the output layer node may include the weight corresponding to each hidden layer node. The bias corresponding to the output layer node may include the bias corresponding to each hidden layer node.
[0165] The linear function can be the activation function of the output layer, which can be expressed as:
[0166]
[0167] For the output of the nth hidden layer node, it can be first multiplied by the weight corresponding to the hidden layer node, then added with a bias, and then calculated using a linear function to obtain the second processed data.
[0168] S73, taking the sum of the n second processed data as the model output torque output by the output layer.
[0169] After respectively calculating and obtaining the second processed data corresponding to each hidden layer node, the second processed data may be added together to obtain the model output torque output by the output layer.
[0170] As a specific example of the present invention, when there are 7 hidden layer nodes, for an output layer node, the output of the node can be expressed as:
[0171]
[0172] in, is the output of the output layer node of the neural network, is the nth input of the output layer node of the neural network, which is calculated by multiplying the output of the nth input layer node by the weight and then adding the bias.
[0173] As a specific example of the present invention, Figure 6 and Figure 7 It is a test diagram of the required torque and the actual torque output by the electronic power steering system in the embodiment of the present invention. It can be seen that the results of the two tests on the vehicle are that the actual torque output by the electronic power steering system is highly consistent with the required torque output by the advanced driver assistance system. It can be seen that the vehicle control method of the embodiment of the present invention can effectively improve the accuracy and reliability of the actual torque output by the electronic power steering system.
[0174] It should be noted that, for the sake of simplicity, the method embodiments are described as a series of action combinations, but those skilled in the art should be aware that the embodiments of the present invention are not limited by the order of the actions described, because according to the embodiments of the present invention, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions involved are not necessarily required by the embodiments of the present invention.
[0175] Reference Figure 8 , shows a structural block diagram of a vehicle control device provided in an embodiment of the present invention, which is applied to a compensation device, the compensation device is arranged in an advanced driver assistance system of a vehicle, the vehicle also includes an electronic power steering system, and specifically may include the following modules:
[0176] The required torque acquisition module 801 is used to obtain the required torque information from the advanced driver assistance system;
[0177] The error compensation torque determination module 802 is used to input the required torque information into the target neural network model preset in the compensation device, and obtain the error compensation torque information output by the target neural network model;
[0178] The torque output module 803 is used to send the error compensation torque information to the electronic power steering system; the electronic power steering system is used to output an actual torque consistent with the required torque information based on the error compensation torque information.
[0179] Optionally, the device further comprises:
[0180] A vehicle speed acquisition module is used to obtain vehicle speed information;
[0181] The error compensation torque determination module comprises:
[0182] The error compensation torque determination submodule is used to input the required torque information and the vehicle speed information into a target neural network model preset in the compensation device, and obtain the error compensation torque information output by the target neural network model.
[0183] Optionally, the device further comprises:
[0184] A training data acquisition module, used to acquire a preset training data set, wherein the training data set includes at least one training sample; the training sample includes input sample information and a standard output torque; the input sample information includes a required torque sample;
[0185] The model training module is used to use the training samples to train the preset neural network model to obtain a trained target neural network model.
[0186] Optionally, the input sample information also includes a vehicle speed sample.
[0187] Optionally, the demand torque samples include steady-state torque samples and / or dynamic torque samples.
[0188] Optionally, the neural network model is provided with model parameters;
[0189] The model training module includes:
[0190] A model output acquisition submodule, used to input the input sample information into the neural network model to obtain the model output torque output by the neural network model;
[0191] A parameter adjustment submodule, used for adjusting the model parameters based on the difference between the model output torque and the standard output torque corresponding to the input sample information;
[0192] A repeated training submodule is used to repeatedly execute the steps of inputting the input sample information into the neural network model, obtaining the model output torque output by the neural network model, and adjusting the model parameters based on the difference between the model output torque and the standard output torque corresponding to the input sample information, until the number of repeated executions reaches a preset training number threshold, or the difference between the model output torque and the standard output torque corresponding to the input sample information is less than a preset error threshold, thereby obtaining a trained target neural network model.
[0193] Optionally, the neural network includes an input layer, a hidden layer connected to the input layer, and an output layer connected to the hidden layer;
[0194] The model parameters include a hidden layer weight matrix and a hidden layer bias matrix corresponding to the hidden layer, and also include an input layer weight matrix and an input layer bias matrix corresponding to the output layer;
[0195] The model output acquisition submodule includes:
[0196] A first data acquisition unit, used for inputting the input sample information into the input layer respectively to obtain first data output by the input layer;
[0197] A second data acquisition unit, configured to input the first data into the hidden layer, and obtain second data calculated by the hidden layer based on the first data, the hidden layer weight matrix, and the hidden layer bias matrix;
[0198] A model output torque acquisition unit is used to input the second data into the output layer to obtain the model output torque calculated by the output layer based on the second data, the output layer weight matrix and the output layer bias matrix.
[0199] Optionally, the input layer includes m input layer nodes, and the hidden layer includes n hidden layer nodes; wherein m≥1, n≥1; the first data includes outputs of the m input layer nodes; the hidden layer weight matrix includes weights corresponding to the hidden layer nodes, and the hidden layer bias matrix includes biases corresponding to the hidden layer nodes;
[0200] The second data acquisition unit includes:
[0201] A first node output acquisition subunit, used for acquiring the outputs of the m input layer nodes respectively for the nth hidden layer node;
[0202] A first data processing subunit is used to process the outputs of the m input layer nodes respectively by using a hyperbolic tangent function, a weight corresponding to the nth hidden layer node, and a bias corresponding to the nth hidden layer node to obtain m first processed data;
[0203] The second data determination subunit is used to use the sum of the m first processed data as the output of the nth hidden layer node, and use the outputs of the n hidden layer nodes as the second data.
[0204] Optionally, the model output torque acquisition unit includes:
[0205] A second node output acquisition subunit, used for respectively acquiring outputs of the n hidden layer nodes for the output layer;
[0206] A second data processing subunit is used to process the outputs of the n hidden layer nodes respectively using a linear function, the output layer weight matrix, and the output layer bias matrix to obtain n second processed data;
[0207] The model output torque determination subunit is used to take the sum of the n second processed data as the model output torque output by the output layer.
[0208] As for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.
[0209] An embodiment of the present invention further provides a vehicle, comprising a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus;
[0210] The memory is used to store computer programs;
[0211] The processor is used to implement the method described in any one of the embodiments of the present invention when executing the program stored in the memory.
[0212] In addition, an embodiment of the present invention further provides an electronic device, such as Fig. 9 As shown, it includes a processor 901, a communication interface 902, a memory 903 and a communication bus 904, wherein the processor 901, the communication interface 902, and the memory 903 communicate with each other through the communication bus 904.
[0213] Memory 903, used for storing computer programs;
[0214] The processor 901 is used to execute the program stored in the memory 903 to implement the following steps:
[0215] Obtaining required torque information from the advanced driver assistance system;
[0216] Inputting the required torque information into a target neural network model preset in the compensation device, and obtaining the error compensation torque information output by the target neural network model;
[0217] The error compensation torque information is sent to the electronic power steering system; the electronic power steering system is used to output an actual torque consistent with the required torque information based on the error compensation torque information.
[0218] Optionally, the method further comprises:
[0219] Get vehicle speed information;
[0220] The step of inputting the required torque information into a target neural network model preset in the compensation device and obtaining the error compensation torque information output by the target neural network model comprises:
[0221] The required torque information and the vehicle speed information are input into a target neural network model preset in the compensation device, and the error compensation torque information output by the target neural network model is obtained.
[0222] Optionally, the method further comprises:
[0223] Acquire a preset training data set, wherein the training data set includes at least one training sample; the training sample includes input sample information and a standard output torque; the input sample information includes a required torque sample;
[0224] The training samples are used to train a preset neural network model to obtain a trained target neural network model.
[0225] Optionally, the input sample information also includes a vehicle speed sample.
[0226] Optionally, the demand torque samples include steady-state torque samples and / or dynamic torque samples.
[0227] Optionally, the neural network model is provided with model parameters;
[0228] The step of using the training samples to train a preset neural network model to obtain a trained target neural network model includes:
[0229] Inputting the input sample information into a neural network model to obtain a model output torque output by the neural network model;
[0230] Adjusting the model parameters based on the difference between the model output torque and the standard output torque corresponding to the input sample information;
[0231] Repeat the steps of inputting the input sample information into the neural network model, obtaining the model output torque output by the neural network model, and adjusting the model parameters based on the difference between the model output torque and the standard output torque corresponding to the input sample information, until the number of repeated executions reaches a preset training number threshold, or the difference between the model output torque and the standard output torque corresponding to the input sample information is less than a preset error threshold, thereby obtaining a trained target neural network model.
[0232] Optionally, the neural network includes an input layer, a hidden layer connected to the input layer, and an output layer connected to the hidden layer;
[0233] The model parameters include a hidden layer weight matrix and a hidden layer bias matrix corresponding to the hidden layer, and also include an input layer weight matrix and an input layer bias matrix corresponding to the output layer;
[0234] The step of inputting the input sample information into the neural network model to obtain the model output torque output by the neural network model includes:
[0235] Inputting the input sample information into the input layer respectively to obtain the first data output by the input layer;
[0236] Inputting the first data into the hidden layer to obtain second data calculated by the hidden layer based on the first data, the hidden layer weight matrix and the hidden layer bias matrix;
[0237] The second data is input into the output layer to obtain the model output torque calculated by the output layer based on the second data, the output layer weight matrix and the output layer bias matrix.
[0238] Optionally, the input layer includes m input layer nodes, and the hidden layer includes n hidden layer nodes; wherein m≥1, n≥1; the first data includes outputs of the m input layer nodes; the hidden layer weight matrix includes weights corresponding to the hidden layer nodes, and the hidden layer bias matrix includes biases corresponding to the hidden layer nodes;
[0239] The step of inputting the first data into the hidden layer to obtain second data calculated by the hidden layer based on the first data, the hidden layer weight matrix and the hidden layer bias matrix comprises:
[0240] For the nth hidden layer node, respectively obtain the outputs of the m input layer nodes;
[0241] Using a hyperbolic tangent function, a weight corresponding to the nth hidden layer node, and a bias corresponding to the nth hidden layer node, the outputs of the m input layer nodes are processed respectively to obtain m first processed data;
[0242] The sum of the m first processed data is used as the output of the nth hidden layer node, and the outputs of the n hidden layer nodes are used as the second data.
[0243] Optionally, the step of inputting the second data into the output layer to obtain a model output torque calculated by the output layer based on the second data, the output layer weight matrix, and the output layer bias matrix includes:
[0244] For the output layer, respectively obtain the outputs of the n hidden layer nodes;
[0245] Using a linear function, the output layer weight matrix, and the output layer bias matrix, the outputs of the n hidden layer nodes are processed respectively to obtain n second processed data;
[0246] The sum of the n second processed data is used as the model output torque output by the output layer.
[0247] The communication bus mentioned in the above terminal can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The communication bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.
[0248] The communication interface is used for communication between the above terminal and other devices.
[0249] The memory may include a random access memory (RAM) or a non-volatile memory, such as at least one disk memory. Optionally, the memory may also be at least one storage device located away from the aforementioned processor.
[0250] The above-mentioned processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.
[0251] In another embodiment provided by the present invention, a computer-readable storage medium is provided, in which instructions are stored. When the computer-readable storage medium is run on a computer, the computer executes the vehicle control method described in the above embodiment.
[0252] In another embodiment of the present invention, a computer program product including instructions is provided. When the computer program product is run on a computer, the computer executes the vehicle control method described in the above embodiment.
[0253] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented by software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present invention is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from one website site, computer, server or data center to another website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrated. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive Solid State Disk (SSD)), etc.
[0254] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.
[0255] Each embodiment in this specification is described in a related manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.
[0256] The above description is only a preferred embodiment of the present invention and is not intended to limit the protection scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention are included in the protection scope of the present invention.
Claims
1. A vehicle control method, characterized in that: The method is applied to a compensation device, the compensation device is arranged in an advanced driver assistance system of a vehicle, the vehicle also includes an electronic power steering system, and the method includes: Obtaining required torque information from the advanced driver assistance system and obtaining vehicle speed information; wherein the vehicle speed information is associated with an actual torque output range of the electronic power steering system; Inputting the required torque information and the vehicle speed information into a target neural network model preset in the compensation device, and obtaining the error compensation torque information output by the target neural network model; The error compensation torque information is sent to the electronic power steering system; the electronic power steering system is used to output an actual torque consistent with the required torque information based on the error compensation torque information to compensate for the error generated by the electronic power steering system.
2. The method according to claim 1, characterized in that The method further comprises: Acquire a preset training data set, wherein the training data set includes at least one training sample; the training sample includes input sample information and a standard output torque; the input sample information includes a required torque sample; The training samples are used to train a preset neural network model to obtain a trained target neural network model.
3. The method according to claim 2, characterized in that The input sample information also includes a vehicle speed sample.
4. The method according to claim 2, characterized in that: The demand torque samples include steady-state torque samples and / or dynamic torque samples.
5. The method according to any one of claims 2 to 4, characterized in that: The neural network model is provided with model parameters; The step of using the training samples to train a preset neural network model to obtain a trained target neural network model includes: Inputting the input sample information into a neural network model to obtain a model output torque output by the neural network model; Adjusting the model parameters based on the difference between the model output torque and the standard output torque corresponding to the input sample information; Repeat the steps of inputting the input sample information into the neural network model, obtaining the model output torque output by the neural network model, and adjusting the model parameters based on the difference between the model output torque and the standard output torque corresponding to the input sample information, until the number of repeated executions reaches a preset training number threshold, or the difference between the model output torque and the standard output torque corresponding to the input sample information is less than a preset error threshold, thereby obtaining a trained target neural network model.
6. The method according to claim 5, characterized in that The neural network includes an input layer, a hidden layer connected to the input layer, and an output layer connected to the hidden layer; The model parameters include a hidden layer weight matrix and a hidden layer bias matrix corresponding to the hidden layer, and also include an input layer weight matrix and an input layer bias matrix corresponding to the output layer; The step of inputting the input sample information into the neural network model to obtain the model output torque output by the neural network model includes: Inputting the input sample information into the input layer respectively to obtain the first data output by the input layer; Inputting the first data into the hidden layer to obtain second data calculated by the hidden layer based on the first data, the hidden layer weight matrix and the hidden layer bias matrix; The second data is input into the output layer to obtain the model output torque calculated by the output layer based on the second data, the output layer weight matrix and the output layer bias matrix.
7. The method according to claim 6, characterized in that The input layer includes m input layer nodes, and the hidden layer includes n hidden layer nodes; wherein m≥1, n≥1; the first data includes outputs of the m input layer nodes; the hidden layer weight matrix includes weights corresponding to the hidden layer nodes, and the hidden layer bias matrix includes biases corresponding to the hidden layer nodes; The step of inputting the first data into the hidden layer to obtain second data calculated by the hidden layer based on the first data, the hidden layer weight matrix and the hidden layer bias matrix comprises: For the nth hidden layer node, respectively obtain the outputs of the m input layer nodes; Using a hyperbolic tangent function, a weight corresponding to the nth hidden layer node, and a bias corresponding to the nth hidden layer node, the outputs of the m input layer nodes are processed respectively to obtain m first processed data; The sum of the m first processed data is used as the output of the nth hidden layer node, and the outputs of the n hidden layer nodes are used as the second data.
8. The method according to claim 7, characterized in that The step of inputting the second data into the output layer to obtain the model output torque calculated by the output layer based on the second data, the output layer weight matrix and the output layer bias matrix includes: For the output layer, respectively obtain the outputs of the n hidden layer nodes; Using a linear function, the output layer weight matrix, and the output layer bias matrix, the outputs of the n hidden layer nodes are processed respectively to obtain n second processed data; The sum of the n second processed data is used as the model output torque output by the output layer.
9. A vehicle, characterized in that: It includes a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus; The memory is used to store computer programs; The processor is used to implement the method according to any one of claims 1 to 8 when executing the program stored in the memory.
10. An electronic device, characterized in that: It includes a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus; The memory is used to store computer programs; The processor is used to implement the method according to any one of claims 1 to 8 when executing the program stored in the memory.
11. A computer-readable medium having instructions stored thereon, which, when executed by one or more processors, cause the processors to perform the method according to any one of claims 1 to 8.
