Vehicle state control method and device, electronic equipment and readable storage medium
By using the vehicle state prediction network and a fuzzy controller in the vehicle control system, the problem in the prior art is difficult to accurately predict the vehicle state in a complex driving environment, and precise adjustment of the vehicle yaw stability control is achieved.
Patent Information
- Application Number
- CN202510251102.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-04
- Publication Date
- 2025-05-16
AI Technical Summary
The prior art is difficult to accurately predict vehicle status in complex and changeable driving environments, resulting in the inability to control the yaw stability of the vehicle.
By obtaining the vehicle state observation measurement of the historical moment of the target vehicle and the vehicle state observation measurement of the current moment of the vehicle, input the vehicle state prediction network for state prediction, calculate the state deviation, and generate control instructions to adjust the vehicle state to achieve stability.
It realizes more accurate vehicle status prediction and real-time control, solves the problem of vehicle yaw stability control, and makes vehicle status control more flexible and accurate.
Smart Images
Figure CN120003501A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of automobile driving technology, and in particular to a vehicle state control method, device, electronic device and readable storage medium. Background Art
[0002] Chassis-by-wire technology refers to the use of electronic signals in the vehicle control system to transmit control instructions to achieve precise control of the vehicle chassis system (such as steering, braking, suspension, etc.) instead of traditional physical mechanical connections.
[0003] Motion control for intelligent driving vehicles through chassis-by-wire technology is one of the core issues in the field of intelligent vehicle research. Existing control schemes such as proportional-integral-derivative control (PID) adjust the control signal through proportional, integral and differential terms to achieve the purpose of stabilizing the vehicle's motion state. However, in complex and changeable driving environments, more intelligent and efficient control algorithms are needed to accurately reflect the actual dynamics of the vehicle. The parameters of the PID controller (proportional coefficient, integral time and differential time) are difficult to adjust to the optimal level, and repeated experiments and debugging are often required. The design of the PID controller is usually based on the linearized model of the system. In nonlinear or time-varying systems, this simplified model may not accurately reflect the actual dynamics of the vehicle, and thus cannot provide accurate vehicle state estimation and prediction, resulting in the inability to achieve stability control of the vehicle's yaw. Summary of the invention
[0004] In view of this, the embodiments of the present application provide a vehicle state control method, device, electronic device and readable storage medium to solve the problem in the prior art that the vehicle yaw stability control cannot be achieved due to the inability to accurately predict the vehicle state.
[0005] According to a first aspect of an embodiment of the present application, a vehicle state control method is provided, comprising:
[0006] Obtain the vehicle state observation value of the target vehicle at the first preset historical moment before the current moment and the vehicle state observation value of the target vehicle at the current moment;
[0007] Inputting the vehicle state observation quantity of the first preset historical moment into the vehicle state prediction network, performing state prediction on the target vehicle based on the vehicle state prediction network, and obtaining the vehicle state prediction quantity of the target vehicle at the current moment;
[0008] Determine the vehicle state deviation between the current vehicle state observation and the current vehicle state prediction, and generate a control instruction corresponding to the vehicle state deviation;
[0009] The vehicle state of the target vehicle at the current moment is adjusted based on the control instruction.
[0010] According to a second aspect of an embodiment of the present application, a vehicle state control device is provided, comprising:
[0011] An acquisition module is configured to acquire a vehicle state observation value of the target vehicle at a first preset historical moment before the current moment and a vehicle state observation value of the target vehicle at the current moment;
[0012] A vehicle state prediction module is configured to input the vehicle state observations at the first preset historical moments into the vehicle state prediction network, perform state prediction on the target vehicle based on the vehicle state prediction network, and obtain a current moment vehicle state prediction of the target vehicle at the current moment;
[0013] A deviation correction module is configured to determine a vehicle state deviation between an observed vehicle state at a current moment and a predicted vehicle state at a current moment, and generate a control instruction corresponding to the vehicle state deviation;
[0014] The control module is configured to adjust the vehicle state of the target vehicle at a current moment based on the control instruction.
[0015] According to a third aspect of an embodiment of the present application, an electronic device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above method when executing the computer program.
[0016] According to a fourth aspect of an embodiment of the present application, a readable storage medium is provided, which stores a computer program, and when the computer program is executed by a processor, the steps of the above method are implemented.
[0017] Compared with the prior art, the embodiments of the present application have the following beneficial effects: by obtaining the vehicle state observations of the target vehicle at the first preset historical moments before the current moment and the vehicle state observations of the target vehicle at the current moment, the vehicle state observations at the first preset historical moments provide the past state trend of the target vehicle, the vehicle state observations at the first preset historical moments can provide the vehicle state prediction network with sufficient data to predict the future state of the target vehicle, and the vehicle state observations at the current moment provide the latest state information of the target vehicle, and can be used to compare with the predicted state to evaluate the accuracy of the prediction and calculate the deviation. The vehicle state observations at the first preset historical moments are input into the vehicle state prediction network, and the state of the target vehicle at the current moment is predicted based on the vehicle state prediction network, the long-term dependency in the time series data (the vehicle state observations at the first preset historical moments) is captured, and the vehicle state prediction quantity at the current moment of the target vehicle at the current moment is obtained. The vehicle state deviation between the observed vehicle state at the current moment and the predicted vehicle state at the current moment is calculated. The vehicle state deviation can reflect the gap between the actual state and the predicted state, and a control instruction corresponding to the vehicle state deviation is generated. Based on the generated control instruction, the vehicle state of the target vehicle at the current moment is adjusted to meet the expected target, so as to achieve the purpose of stabilizing the vehicle. The vehicle state control method proposed in this application can more accurately predict the state of the vehicle, and adjust the control strategy according to the vehicle state deviation calculated in real time, so as to solve the problem in the prior art that the vehicle state cannot be accurately predicted and thus the vehicle yaw stability control cannot be achieved, making the vehicle state control more flexible and accurate. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.
[0019] Figure 1 It is a flow chart of a vehicle state control method provided in an embodiment of the present application;
[0020] Figure 2 It is a flowchart of another vehicle state prediction method provided in an embodiment of the present application;
[0021] Figure 3 It is a structural diagram of a vehicle state prediction network provided in an embodiment of the present application;
[0022] Figure 4 is a structural schematic diagram of a vehicle state control device provided in an embodiment of the present application;
[0023] Figure 5 It is a structural schematic diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0024] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present application.
[0025] A vehicle state control method and device according to an embodiment of the present application will be described in detail below with reference to the accompanying drawings.
[0026] Figure 1 is a flow chart of a vehicle state control method provided by an embodiment of the present application, such as Figure 1 As shown, the vehicle state control method includes:
[0027] Step 101, obtaining the vehicle state observation value of the target vehicle at the first preset historical moments before the current moment and the vehicle state observation value of the target vehicle at the current moment.
[0028] Specifically, the initial observation of the vehicle state can be obtained through vehicle sensors and control systems such as wheel speed sensors, gyroscopes and inertial measurement units installed on the target vehicle. The initial observation of the vehicle state may include but is not limited to information such as the vehicle observed position coordinates, the vehicle observed longitudinal velocity (vx), and the vehicle observed longitudinal acceleration (ax). The initial observation of the vehicle state is normalized and scaled to a standard range (such as 0 to 1) to obtain the vehicle state observation.
[0029] It can be understood that the first preset historical moment vehicle state observation is the vehicle state data of the first preset number of consecutive moments traced back from a certain moment during the driving process of the target vehicle, which can reflect the movement state of the target vehicle in the past period of time. The first preset historical moment vehicle state observation provides dynamic information of the target vehicle in the past period of time, which can reflect the behavior pattern and trend of the target vehicle, which is highly referenced for predicting the future state of the target vehicle. By using the first preset historical moment vehicle state observation, the vehicle state prediction network can obtain more contextual information, thereby improving the accuracy and reliability of vehicle state prediction.
[0030] The current moment vehicle state observation is the vehicle state data of the target vehicle at a specific moment (i.e., the current moment), which can reflect the actual state of the vehicle at the current moment. The current moment vehicle state observation will be used to compare with the current moment vehicle state prediction output by the vehicle state prediction network to obtain the deviation between the two, and input these deviation information into the fuzzy controller. The fuzzy controller dynamically adjusts the control instructions according to the deviation, so that the state of the target vehicle can be adjusted to the ideal state in time, and dynamically adjusts and controls the vehicle state.
[0031] Specifically, as an example, It can be the vehicle state observation value of the first preset historical moment, is the vehicle state observation at the current moment.
[0032] Step 102, inputting the vehicle state observations at the first preset historical moments into the vehicle state prediction network, performing state prediction on the target vehicle based on the vehicle state prediction network, and obtaining the current moment vehicle state prediction of the target vehicle at the current moment.
[0033] Specifically, the vehicle state prediction network can be obtained by training a long short-term memory network (Long Short-Term Memory, LSTM) based on a training sample set.
[0034] Specifically, the training process of the vehicle state prediction network can be as follows: obtain multiple groups of vehicle state data, and use ten-fold cross validation to randomly distribute the multiple groups of vehicle state data into training sample sets and test sample sets. The training sample set includes multiple training samples, and each training sample includes vehicle state observations at multiple historical moments of the vehicle and vehicle state observations at multiple future moments after multiple historical moments. Use the training sample set to train LSTM. The input of LSTM is the vehicle state observations at multiple historical moments, and the output is the vehicle state predictions at multiple future moments for the related inputs. The optimizer selected for LSTM training can be Adam. Adam shows a fast convergence speed in many tasks, is not very sensitive to the choice of hyperparameters, is suitable for processing large-scale data and high-dimensional parameter space, and can handle sparse gradients and non-stationary targets. The hyperparameter settings for training are shown in Table 1:
[0035] Table 1 Training parameter settings
[0036]
[0037]
[0038] The resulting performance of the model is evaluated using the root mean square error RMSE:
[0039]
[0040] Among them, k represents the prediction step length; represent the true value and the predicted value respectively.
[0041] The trained LSTM (i.e., vehicle state prediction network) can predict the vehicle state prediction at a future moment through the vehicle state observation at a historical moment.
[0042] LSTM can effectively retain and forget information through its internal memory units and gating mechanisms (input gate, forget gate, output gate), thereby capturing long-term dependencies in the input sequence. Even if the time series is long, it can remember important information fragments, so the vehicle state prediction network obtained by training LSTM based on the training sample set also has this feature. Through the training process, the long-term rules in the training sample set are learned. The vehicle state prediction network can handle the long-term dependencies in the time series data, thereby more accurately predicting the future vehicle state. Even in a complex driving environment, the vehicle state prediction network can provide relatively accurate prediction results. In addition, the vehicle state prediction network has high computational efficiency in prediction and control, which can meet the requirements of real-time control. The vehicle state prediction network algorithm can be processed in parallel on hardware accelerators such as GPUs to improve real-time performance. The vehicle state prediction network can meet the requirements of real-time control when processing complex dynamic systems through its parallel computing capabilities, efficient time series processing, adaptive learning and hardware acceleration.
[0043] The vehicle state observations at the first preset historical moment are input into the vehicle state prediction network. Based on the trained vehicle state prediction network, the vehicle state observations at the first preset historical moment are processed using the learned rules to predict the state of the target vehicle at the current moment and to predict the vehicle state prediction value of the target vehicle at the current moment.
[0044] The vehicle state prediction of the target vehicle at the current moment includes parameters such as vehicle position coordinates, vehicle longitudinal velocity, and vehicle longitudinal acceleration. By using observations at multiple historical moments, the vehicle state prediction network can capture the trend and pattern of the evolution of the vehicle state over time, and obtain accurate prediction results, that is, the vehicle state prediction at the current moment. This method is particularly effective in nonlinear and time-varying systems. After predicting the vehicle state prediction at the current moment, the vehicle state prediction at the current moment can be compared with the vehicle state observation at the current moment, and then control instructions can be generated to achieve real-time control of the target vehicle. By continuously acquiring and predicting the vehicle state observation, the control strategy can be adjusted in real time to respond to any changes in the state of the target vehicle, respond more flexibly to different driving conditions and driving environments, and improve the safety and handling performance of the target vehicle.
[0045] Step 103, determining the vehicle state deviation between the current vehicle state observation and the current vehicle state prediction, and generating a control instruction corresponding to the vehicle state deviation.
[0046] In some embodiments, in vehicle state control, fuzzy control can generate corresponding control signals based on the vehicle state prediction quantity at the current moment predicted by the vehicle state prediction network and the vehicle state observation quantity at the current moment. By designing a fuzzy controller, the control strategy can be quickly adjusted according to the real-time state information and prediction results to adapt to different driving environments and road conditions, so as to achieve accurate control of the vehicle state. The design of the fuzzy controller can include steps such as defining fuzzy variables, formulating fuzzy rules, and performing fuzzy reasoning and defuzzification. Specifically, it can include:
[0047] Define the input and output fuzzy variables. The input variables include the vehicle position coordinate error (x_error, y_error), vehicle speed error (vx_error), and vehicle acceleration error (ax_error). The output variables are control instructions (such as steering angle, acceleration, etc.). Membership functions can be used to quantify the degree to which a specific input belongs to the fuzzy set. The precise input value (input variable) is converted into a membership degree between 0 and 1. The membership function can be a triangular, trapezoidal, or Gaussian function.
[0048] Define fuzzy rules, that is, define fuzzy rules based on expert experience or experimental data to describe the relationship between input variables and output variables. Fuzzy rules are statements built in an "if-then" format that describe the logical relationship between input variables and output variables, for example, "If x_error is low and y_error is low, then the control signal is small." This rule means that if both coordinate errors are small, a smaller control adjustment should be applied. The defined fuzzy rules help the fuzzy controller determine what control output should be produced under given input conditions;
[0049] Fuzzy reasoning and defuzzification, fuzzy reasoning is to reason about input variables according to defined fuzzy rules to generate output variables corresponding to the input variables. This step involves mapping the input variables to fuzzy sets and determining the output variables according to the fuzzy rules; defuzzification is to convert the aggregated fuzzy output (i.e., output variable) into a single precise value that can be used as a control signal. Common defuzzification methods include the centroid method (finding the centroid of the fuzzy set) and the maximum membership method (selecting the output with the highest membership).
[0050] In addition, the vehicle's yaw rate and center of mass sideslip angle are important parameters that affect vehicle stability. To ensure the driving stability of the target vehicle, these two parameters need to be limited within the desired range. In the process of designing the above-mentioned fuzzy controller, the vehicle's yaw rate and center of mass sideslip angle can be used as optimization constraints to achieve the purpose of adjusting the speed and acceleration of the target vehicle, which can ensure that the target vehicle remains stable during driving and avoid dangerous situations such as sideslipping and loss of control.
[0051] Specifically, by comparing the observed vehicle state at the current moment with the predicted vehicle state at the current moment, the vehicle state deviation can be obtained. The vehicle state deviation reflects the difference between the actual state of the target vehicle at the current moment and the predicted state, and is the basis for subsequent control. The vehicle state deviation is input into the fuzzy controller, and the fuzzy processor generates a control instruction corresponding to the vehicle state deviation based on the vehicle state deviation. Among them, the fuzzy controller can process the above-mentioned inaccurate or fuzzy vehicle state deviation information, and generate control instructions corresponding to the vehicle state deviation according to preset rules and logic, such as adjusting the steering angle, acceleration, etc., to correct the vehicle state deviation of the target vehicle and realize the control of the vehicle yaw stability. Compared with the traditional control scheme, which may require complex filtering and compensation measures when facing noise and uncertainty, the control method based on the fuzzy controller is more flexible and robust.
[0052] Step 104: adjusting the vehicle state of the target vehicle at the current moment based on the control instruction.
[0053] Specifically, the control instructions act on the target vehicle through the target vehicle's actuators (such as the throttle system, steering system, etc.), and adjust the steering angle, acceleration, etc. of the target vehicle in real time, so that the vehicle state of the target vehicle is closer to the predicted vehicle state. This helps to achieve precise control of the target vehicle, especially in complex driving environments. It can also significantly reduce fluctuations and deviations during driving, improve driving stability, bring a smoother driving experience, and reduce passenger discomfort.
[0054] Based on the vehicle state control method proposed in this application, the vehicle state prediction network is obtained by training the LSTM for the vehicle state observation at the first preset historical moment, and the state of the target vehicle at the current moment is predicted based on the vehicle state prediction network, and the long-term dependency in the time series data (the vehicle state observation at the first preset historical moment) is captured to obtain the vehicle state prediction quantity at the current moment of the target vehicle. The vehicle state deviation between the vehicle state observation at the current moment and the vehicle state prediction quantity at the current moment is calculated. The vehicle state deviation can reflect the gap between the actual state and the predicted state, and the vehicle state deviation is input into the fuzzy controller. The fuzzy controller generates a control instruction corresponding to the vehicle state deviation according to the size and direction of the vehicle state deviation. Based on the generated control instruction, the vehicle state of the target vehicle at the current moment is adjusted to meet the expected goal, so as to achieve the purpose of stabilizing the vehicle. The vehicle state control method proposed in this application adjusts the control strategy according to the vehicle state deviation calculated in real time, solves the problem that the vehicle yaw stability control cannot be realized due to the inability to accurately predict the vehicle state in the prior art, and makes the vehicle state control more flexible and accurate.
[0055] In some embodiments, the state of the target vehicle can also be predicted based on the vehicle state prediction network to obtain the vehicle state prediction value of the target vehicle at a second preset future moment after the current moment; the target parameters in the vehicle state prediction value at the current moment and the vehicle state prediction value at the second preset future moment are identified, and when the value of the target parameter is greater than the preset threshold value corresponding to the target parameter, abnormal state warning information is generated.
[0056] Specifically, the vehicle state prediction network can predict the vehicle state of the vehicle at a future moment based on the input vehicle state at a historical moment. The first preset vehicle state observation at a historical moment (vehicle state at a historical moment) is input into the vehicle state prediction network, and the vehicle state prediction network processes the first preset vehicle state observation at a historical moment, captures and memorizes important information in a long time series, generates a vehicle state prediction at the current moment and a vehicle state prediction at a second preset future moment (vehicle state at a future moment), and understands the possible behavior and state changes of the target vehicle in advance, providing a basis for subsequent abnormality detection.
[0057] The target parameters in the predicted vehicle state at the current moment and the predicted vehicle state at the second preset future moment may be vehicle speed, vehicle acceleration, etc. The target parameters in the predicted values at the current moment and the future moment are identified, and the predicted values are compared with the preset thresholds to detect whether there is an abnormal state. When the value of the target parameter is greater than the preset threshold, the abnormal state is identified, abnormal state warning information is generated, and the driver or automatic driving system is notified in time so that appropriate measures can be taken to avoid potential dangerous situations.
[0058] In some embodiments, the state of the target vehicle is predicted based on the vehicle state prediction network to obtain the predicted vehicle state of the target vehicle at a second preset future time after the current time, including:
[0059] Taking the first preset historical moment vehicle state observation as an input sequence, inputting the input sequence into the vehicle state prediction network, and obtaining the current moment vehicle state prediction of the target vehicle at the current moment;
[0060] The current moment vehicle state prediction value is used to replace the historical moment vehicle state observation value with the earliest timestamp in the input sequence to obtain an updated input sequence, and the updated input sequence is input into the vehicle state prediction network to obtain the next future moment vehicle state prediction value of the current moment vehicle state observation value of the target vehicle;
[0061] According to the preset prediction step length, the above steps are repeated until the required vehicle state prediction value at the second preset future moment is obtained, wherein the preset prediction step length corresponds to the vehicle state prediction value at the second preset future moment.
[0062] In some embodiments, reference Figure 2 Vehicle state prediction flow chart, is the vehicle state observation at the first preset historical moment, is the predicted value of the vehicle state at the current moment. In the prediction stage, the vehicle state at the future moment is predicted, and a sliding window is set for this purpose, with a window size of p.
[0063] In the first step, the vehicle state observations at the first preset historical moment (i.e. ) as the input sequence, and input the input sequence into the vehicle state prediction network to obtain the current vehicle state prediction value of the target vehicle at the current time After the first step of prediction is completed, in the second step, the current vehicle state prediction Add to the sliding window and remove the oldest (i.e., the vehicle state observation at the earliest historical moment in the above input sequence), and the updated input sequence (i.e. ), and update the input sequence (i.e. ) Input the vehicle state prediction network to obtain the current vehicle state observation of the target vehicle The vehicle state prediction at the next future moment
[0064] After the second step prediction is completed, in the third step, Add to the sliding window and remove The earliest Get the updated input sequence (i.e. ), and update the input sequence (i.e. ) Input the vehicle state prediction network and get The vehicle state prediction at the next future moment According to the preset prediction step size (i.e., the number of future moments to be predicted, such as the second preset number), the above steps are repeated. After each prediction, the sliding window content is updated, and the vehicle state observation value at the earliest historical moment with the latest prediction value is replaced, so that the data in the sliding window always reflects the most recent time series information, until the required vehicle state prediction value at the second preset future moment (i.e., ) .
[0065] In some embodiments, the current moment vehicle state observation quantity includes the current moment vehicle observation position coordinates, the current moment vehicle observation speed, and the current moment vehicle observation acceleration, and the current moment vehicle state prediction quantity includes the current moment vehicle prediction position coordinates, the current moment vehicle prediction speed, and the current moment vehicle prediction acceleration;
[0066] Determine the vehicle state deviation between the current vehicle state observation and the current vehicle state prediction, including:
[0067] Calculate the deviation between the current vehicle observation position coordinates and the current vehicle prediction position coordinates to obtain the vehicle position coordinate deviation;
[0068] Calculate the deviation between the current vehicle observed speed and the current vehicle predicted speed to obtain the vehicle speed deviation;
[0069] Calculate the deviation between the current observed acceleration of the vehicle and the current predicted acceleration of the vehicle to obtain the vehicle acceleration deviation;
[0070] The vehicle position coordinate deviation, the vehicle speed deviation and the vehicle acceleration deviation are determined as the vehicle state deviation.
[0071] Specifically, the current moment vehicle state observation is the vehicle state data acquired in real time through sensors or other technical means, including the current moment vehicle observation position coordinates, the current moment vehicle observation speed, and the current moment vehicle observation acceleration. The current moment vehicle state observation is a direct reflection of the current actual state of the target vehicle. The current moment vehicle state prediction is the state prediction result of the future state of the vehicle based on the vehicle state prediction network, which is an estimate of the future state of the vehicle based on historical vehicle state data, including the current moment vehicle prediction position coordinates, the current moment vehicle prediction speed, and the current moment vehicle prediction acceleration. Calculate the deviation between the current moment vehicle observation position coordinates and the current moment vehicle prediction position coordinates. By comparing the observed position and the predicted position, the vehicle position coordinate deviation is obtained. The vehicle position coordinate deviation can reflect the difference between the actual position and the predicted position of the target vehicle, and understand whether the target vehicle is traveling along the expected path. If the vehicle position coordinate deviation is large, the control command needs to be adjusted to return the target vehicle to the correct path.
[0072] Calculate the deviation between the current vehicle observation speed and the current vehicle prediction speed. The current vehicle observation speed is the longitudinal observation speed of the target vehicle, and the current vehicle prediction speed is the longitudinal prediction speed of the target vehicle. By comparing the observation speed and the prediction speed, the vehicle speed deviation is obtained. The vehicle speed deviation can reflect the deviation between the actual motion state and the predicted motion state of the target vehicle. If the vehicle speed deviation is large, the speed of the target vehicle needs to be adjusted to make it close to the expected speed (predicted speed).
[0073] Calculate the deviation between the current vehicle observed acceleration and the current vehicle predicted acceleration. The current vehicle observed acceleration is the longitudinal observed acceleration of the target vehicle, and the current vehicle predicted acceleration is the longitudinal predicted acceleration of the target vehicle. By comparing the observed acceleration and the predicted acceleration, the vehicle acceleration deviation is obtained. The vehicle acceleration deviation can reflect the prediction accuracy of the dynamic characteristics of the target vehicle. If the vehicle acceleration deviation is large, the acceleration of the target vehicle needs to be adjusted to make it close to the expected acceleration (predicted acceleration).
[0074] The vehicle position coordinate deviation, vehicle speed deviation and vehicle acceleration deviation are combined to determine the vehicle state deviation. By combining the deviations in the three aspects, a comprehensive vehicle state deviation is obtained, which can more accurately reflect the gap between the current state of the vehicle and the expected state. It is used to guide subsequent control strategies and guide the target vehicle to make more precise vehicle state adjustments. The above errors are compensated by real-time adjustments, so that the target vehicle can always remain in a stable vehicle state.
[0075] In some embodiments, generating a control instruction corresponding to a vehicle state deviation includes:
[0076] The membership function is used to fuzzify the vehicle state deviation and generate the corresponding input fuzzy set;
[0077] Reasoning the input fuzzy set according to preset rules to generate the output fuzzy set corresponding to the input fuzzy set;
[0078] The output fuzzy set is defuzzified to generate the corresponding control instructions.
[0079] In some embodiments, the vehicle state deviation includes the vehicle position coordinate deviation, the vehicle speed deviation and the vehicle acceleration deviation, which are determined as the vehicle state deviation. The vehicle state deviation is fuzzified using a membership function, and the precise input value (i.e., the vehicle position coordinate deviation, the vehicle speed deviation and the vehicle acceleration deviation) is converted into a membership degree between 0 and 1 to generate a corresponding input fuzzy set.
[0080] Fuzzy processing can handle uncertainty and noise in vehicle state data and has strong robustness. The membership function defines the range and membership degree of each fuzzy set, so that the fuzzy controller can handle continuously changing vehicle state deviations.
[0081] Specifically, fuzzy sets "small", "medium" and "large" can be defined, and the vehicle position coordinate deviation, vehicle speed deviation and vehicle acceleration deviation can be respectively mapped to the fuzzy sets through membership functions.
[0082] According to the input fuzzy set, the corresponding output fuzzy set is generated by using preset rules. The preset rules are usually expressed in the form of "if...then...", describing the relationship between input and output. For example, "if the vehicle position coordinate deviation is large and the vehicle speed deviation error is large, then the control signal is large". The output fuzzy set is defuzzified, and the fuzzy output obtained by reasoning is converted into actual numerical control instructions so that the actuator can use it directly. The defuzzification method can be the centroid method and the maximum membership method.
[0083] By converting the vehicle state deviation into a fuzzy set and performing fuzzy reasoning according to preset rules, and finally generating precise control instructions, the vehicle state can be precisely controlled. Not only can the prediction accuracy be improved through the vehicle state prediction network, but also the prediction error can be dealt with through real-time adjustment to ensure that the target vehicle can maintain good stability under various conditions. The combination of the vehicle state prediction network and fuzzy logic can significantly improve the performance of the vehicle control system, so that the target vehicle can still maintain good stability and safety in complex driving environments.
[0084] In some embodiments, the control instruction includes a throttle control instruction and a steering control instruction, and adjusting the vehicle state of the target vehicle at the current moment based on the control instruction includes:
[0085] The throttle system of the target vehicle is controlled based on the throttle control instruction so that the vehicle speed deviation of the target vehicle is less than a first preset threshold value and the vehicle acceleration deviation is less than a second preset threshold value; the vehicle speed deviation is the deviation between the vehicle observed speed at the current moment in the vehicle state observation quantity at the current moment and the vehicle predicted speed at the current moment in the vehicle state prediction quantity at the current moment, and the vehicle acceleration deviation is the deviation between the vehicle observed acceleration at the current moment in the vehicle state observation quantity at the current moment and the vehicle predicted acceleration at the current moment in the vehicle state prediction quantity at the current moment;
[0086] The steering system of the target vehicle is controlled based on the steering control instruction so that the vehicle position coordinate deviation of the target vehicle is less than a third preset threshold value; the vehicle position coordinate deviation is the deviation between the vehicle observed position coordinate at the current moment in the vehicle state observation quantity at the current moment and the vehicle predicted position coordinate at the current moment in the vehicle state prediction quantity at the current moment.
[0087] Specifically, the throttle control instruction and the steering control instruction are both specific numerical control instructions.
[0088] According to the throttle control command, the throttle system of the target vehicle is adjusted, and the position of the throttle pedal or the throttle opening of the engine is adjusted to control the speed and acceleration of the target vehicle, so that the vehicle speed deviation and the vehicle acceleration deviation of the target vehicle are controlled within an acceptable range, that is, the vehicle speed deviation is less than the first preset threshold value and the vehicle acceleration deviation is less than the second preset threshold value, so that the speed and acceleration of the target vehicle are maintained in the expected state, thereby avoiding instability caused by sudden changes in speed or acceleration. In addition, the vehicle speed deviation and the vehicle acceleration deviation are monitored in real time, and the throttle control command is adjusted according to the feedback to ensure that the vehicle speed deviation and the vehicle acceleration deviation are always within an acceptable range.
[0089] Specifically, the steering system of the target vehicle is adjusted according to the steering control instruction, and the steering angle is adjusted to correct the trajectory coordinate error to control the position coordinate of the target vehicle, so that the position coordinate deviation of the target vehicle is controlled within an acceptable range, that is, the vehicle position coordinate deviation is less than the third preset threshold value, so that the position of the target vehicle can be kept consistent with the expected path and avoid deviation from the predetermined driving route. In addition, the vehicle position coordinate deviation is monitored in real time, and the steering control instruction is adjusted according to the feedback to ensure that the vehicle position coordinate deviation is always within an acceptable range.
[0090] In some embodiments, the vehicle state observations at the first preset historical moments are input into the vehicle state prediction network, and the state of the target vehicle is predicted based on the vehicle state prediction network to obtain the vehicle state prediction at the current moment of the target vehicle, including:
[0091] The input sequence is sequentially input into the input layer of the vehicle state prediction network, and the output sequence of the input layer is sequentially input into the hidden layer of the vehicle state prediction network; the input sequence is the vehicle state observation at the first preset historical moment;
[0092] Based on the connection weights and thresholds between each gating unit and cell unit in the hidden layer, the features of the output sequence of the input layer are abstracted into a new dimensional space to extract the spatiotemporal variation features of the vehicle state of the target vehicle, and the spatiotemporal variation features of the vehicle state of the target vehicle are linearly divided, and the results of the linear division calculated by the hidden layer are input into the fully connected layer. Each gating unit includes an input gate, a forget gate, and an output gate;
[0093] Perform nonlinear transformation on the input value of the fully connected layer to obtain the output value of the fully connected layer of preset dimension;
[0094] The output value of the fully connected layer is input into the regression output layer to predict the vehicle state based on the output value of the fully connected layer, and the predicted vehicle state of the target vehicle at the current moment is obtained.
[0095] In some embodiments, the structure of the vehicle state prediction network is as follows: Figure 3 As shown, the vehicle state prediction network 301 includes: an input layer 302, a hidden layer 303, a fully connected layer 304 and a regression output layer 305.
[0096] An input sequence consisting of the first preset historical moment vehicle state observations is input into the input layer 302 of the vehicle state prediction network 301. The input layer 302 receives the input sequence and passes the first preset historical moment vehicle state observations as input data into the vehicle state prediction network 301 to provide basic data for subsequent processing. The historical moment vehicle state observations include the historical moment vehicle observation position coordinates, the historical moment vehicle observation speed, and the historical moment vehicle observation acceleration.
[0097] Based on the input gate, forget gate, output gate and other gating units in the hidden layer 303, as well as the connection weights and threshold calculations between the cell units, the output sequence features of the input layer are abstracted to a new dimensional space, and the cell units in the hidden layer selectively retain or forget the input information through the gating mechanism, and the features of the input sequence are abstracted to a new dimensional space to extract the spatiotemporal change characteristics of the state of the target vehicle and capture the change pattern of the vehicle state over time. The various gating units and cell units in the hidden layer enable the vehicle state prediction network to selectively store, delete or read information, ignore irrelevant information, and retain important features, thereby improving the accuracy of the prediction and overcoming the gradient vanishing or gradient exploding problems that are prone to occur in traditional recurrent neural networks (RNNs) when processing long sequence data.
[0098] Specifically, the input gate can determine which output sequences of the input layer need to be stored in the memory unit through the sigmoid function, and generate the state of the candidate memory unit through the tanh function. The forget gate determines which information needs to be deleted from the memory unit through the sigmoid function, thereby adjusting the state of the memory unit. The output gate determines which information needs to be output from the memory unit through the sigmoid function, and generates the state of the memory unit through the tanh function, and then multiplies the two to obtain the final output. The memory unit can store information for a long time and will not be easily lost even in long time series, so that the vehicle state prediction network can retain important historical information. The extracted spatiotemporal variation features of the target vehicle state are linearly divided, and the divided results are input into the fully connected layer 304. Through linear division, complex features can be mapped to a lower dimensional space so that the subsequent fully connected layer can better process these features, which helps to simplify feature expression while retaining important information.
[0099] The fully connected layer 304 performs a nonlinear transformation on the input value of the fully connected layer through a nonlinear activation function (such as ReLU, tanh, etc.) to capture more complex feature relationships, which helps to improve the expression ability and prediction accuracy of the vehicle state prediction network and obtain the output value of the fully connected layer of a preset dimension.
[0100] The output value of the fully connected layer is input into the regression output layer 305, and the vehicle state is predicted based on the output value of the fully connected layer to obtain the current moment vehicle state prediction of the target vehicle at the current moment, thereby realizing the mapping from features to prediction values. The current moment vehicle state prediction includes the current moment vehicle prediction position coordinates, the current moment vehicle prediction speed, and the current moment vehicle prediction acceleration.
[0101] By inputting the vehicle state observations at the first preset historical moment into the vehicle state prediction network and processing them through the input layer, hidden layer, fully connected layer and regression output layer, accurate prediction of the vehicle state can be achieved. The above steps work together to enable the vehicle state prediction network to capture the spatiotemporal variation characteristics of the vehicle state and accurately predict the vehicle state based on the spatiotemporal variation characteristics.
[0102] All the above optional technical solutions can be arbitrarily combined to form optional embodiments of the present application, which will not be described one by one here.
[0103] The following is an embodiment of the device of the present application, which can be used to execute the embodiment of the method of the present application. For details not disclosed in the embodiment of the device of the present application, please refer to the embodiment of the method of the present application.
[0104] Figure 4 Schematic diagram of a vehicle state control device provided in an embodiment of the present application. Figure 4 As shown, the vehicle state control device includes:
[0105] An acquisition module 401 is configured to acquire a vehicle state observation value of the target vehicle at a first preset historical moment before the current moment and a vehicle state observation value of the target vehicle at the current moment;
[0106] The vehicle state prediction module 402 is configured to input the vehicle state observations at the first preset historical moments into a vehicle state prediction network, perform state prediction on the target vehicle based on the vehicle state prediction network, and obtain the vehicle state prediction of the target vehicle at the current moment, wherein the vehicle state prediction network is obtained by training the LSTM based on the training sample set;
[0107] The deviation correction module 403 is configured to determine the vehicle state deviation between the vehicle state observation at the current moment and the vehicle state prediction at the current moment, and generate a control instruction corresponding to the vehicle state deviation;
[0108] The control module 404 is configured to adjust the vehicle state of the target vehicle at the current moment based on the control instruction.
[0109] In some embodiments, the vehicle state prediction module 402 is also configured to perform state prediction on the target vehicle based on the vehicle state prediction network to obtain a predicted vehicle state of the target vehicle at a second preset future moment after the current moment; identify target parameters in the predicted vehicle state at the current moment and the predicted vehicle state at the second preset future moment, and generate abnormal state warning information when the value of the target parameter is greater than a preset threshold value corresponding to the target parameter.
[0110] In some embodiments, the vehicle state prediction module 402 is configured to take the vehicle state observations at the first preset historical moment as an input sequence, input the input sequence into the vehicle state prediction network, and obtain the current moment vehicle state prediction of the target vehicle at the current moment; use the current moment vehicle state prediction to replace the historical moment vehicle state observation with the earliest timestamp in the input sequence to obtain an updated input sequence, and input the updated input sequence into the vehicle state prediction network to obtain the next future moment vehicle state prediction of the current moment vehicle state observation of the target vehicle; repeat the above steps according to the preset prediction step size until the required second preset future moment vehicle state prediction is obtained, wherein the preset prediction step size corresponds to the second preset future moment vehicle state prediction.
[0111] In some embodiments, the observed vehicle state at the current moment includes the observed vehicle position coordinates at the current moment, the observed vehicle speed at the current moment, and the observed vehicle acceleration at the current moment; the predicted vehicle state at the current moment includes the predicted vehicle position coordinates at the current moment, the predicted vehicle speed at the current moment, and the predicted vehicle acceleration at the current moment; the deviation correction module 403 is configured to calculate the deviation between the observed vehicle position coordinates at the current moment and the predicted vehicle position coordinates at the current moment to obtain the vehicle position coordinate deviation; calculate the deviation between the observed vehicle speed at the current moment and the predicted vehicle speed at the current moment to obtain the vehicle speed deviation; calculate the deviation between the observed vehicle acceleration at the current moment and the predicted vehicle acceleration at the current moment to obtain the vehicle acceleration deviation; determine the vehicle position coordinate deviation, vehicle speed deviation and vehicle acceleration deviation as the vehicle state deviation.
[0112] In some embodiments, the deviation correction module 403 is configured to use a membership function to fuzzify the vehicle state deviation to generate a corresponding input fuzzy set; infer the input fuzzy set according to preset rules to generate an output fuzzy set corresponding to the input fuzzy set; defuzzify the output fuzzy set to generate a corresponding control instruction.
[0113] In some embodiments, the control instruction includes a throttle control instruction and a steering control instruction. The control module 404 is configured to control the throttle system of the target vehicle based on the throttle control instruction so that the vehicle speed deviation of the target vehicle is less than a first preset threshold and the vehicle acceleration deviation is less than a second preset threshold; the vehicle speed deviation is the deviation between the observed vehicle speed at the current moment in the vehicle state observation quantity at the current moment and the predicted vehicle speed at the current moment in the vehicle state prediction quantity at the current moment, and the vehicle acceleration deviation is the deviation between the observed vehicle acceleration at the current moment in the vehicle state observation quantity at the current moment and the predicted vehicle acceleration at the current moment in the vehicle state prediction quantity at the current moment; the steering system of the target vehicle is controlled based on the steering control instruction so that the vehicle position coordinate deviation of the target vehicle is less than a third preset threshold; the vehicle position coordinate deviation is the deviation between the observed vehicle position coordinates at the current moment in the vehicle state observation quantity at the current moment and the predicted vehicle position coordinates at the current moment in the vehicle state prediction quantity at the current moment.
[0114] In some embodiments, the vehicle state prediction module 402 is configured to input the input sequence into the input layer of the vehicle state prediction network in sequence, and input the output sequence of the input layer into the hidden layer of the vehicle state prediction network in sequence; the input sequence is the vehicle state observation at the first preset historical moment; based on the connection weights and thresholds between each gating unit and cell unit in the hidden layer, the characteristics of the output sequence of the input layer are abstracted to a new dimensional space to extract the spatiotemporal variation characteristics of the vehicle state of the target vehicle, and the spatiotemporal variation characteristics of the vehicle state of the target vehicle are linearly divided, and the result of the linear division calculated by the hidden layer is input into the fully connected layer, and each gating unit includes an input gate, a forgetting gate and an output gate; the input value of the fully connected layer is nonlinearly transformed to obtain the output value of the fully connected layer of a preset dimension; the output value of the fully connected layer is input into the regression output layer to predict the vehicle state based on the output value of the fully connected layer, and obtain the predicted value of the vehicle state of the target vehicle at the current moment.
[0115] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.
[0116] Figure 5 Schematic diagram of an electronic device 5 provided in an embodiment of the present application. Figure 5 As shown, the electronic device 5 of this embodiment includes: a processor 501, a memory 502, and a computer program 503 stored in the memory 502 and executable on the processor 501. When the processor 501 executes the computer program 503, the steps in the above-mentioned method embodiments are implemented. Alternatively, when the processor 501 executes the computer program 503, the functions of the modules / units in the above-mentioned device embodiments are implemented.
[0117] The electronic device 5 may be a desktop computer, a notebook, a PDA, a cloud server, or other electronic device. The electronic device 5 may include, but is not limited to, a processor 501 and a memory 502. Those skilled in the art will appreciate that Figure 5 The electronic device 5 is merely an example and does not limit the electronic device 5 , and may include more or less components than those shown in the figure, or different components.
[0118] The processor 501 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.
[0119] The memory 502 may be an internal storage unit of the electronic device 5, for example, a hard disk or memory of the electronic device 5. The memory 502 may also be an external storage device of the electronic device 5, for example, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the electronic device 5. The memory 502 may also include both an internal storage unit of the electronic device 5 and an external storage device. The memory 502 is used to store computer programs and other programs and data required by the electronic device.
[0120] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units.
[0121] If the integrated module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium (e.g., a computer-readable storage medium). Based on this understanding, the present application implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. The computer program may include computer program code, which may be in source code form, object code form, executable file or some intermediate form. Computer-readable storage media may include: any entity or device capable of carrying computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium, etc.
[0122] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.
Claims
1. A vehicle state control method, characterized in that: include: Obtaining the vehicle state observation value of the target vehicle at the first preset historical moment before the current moment and the vehicle state observation value of the target vehicle at the current moment; Inputting the first preset historical moment vehicle state observations into a vehicle state prediction network, performing state prediction on the target vehicle based on the vehicle state prediction network, and obtaining a current moment vehicle state prediction of the target vehicle at the current moment; Determine a vehicle state deviation between the vehicle state observation at the current moment and the vehicle state prediction at the current moment, and generate a control instruction corresponding to the vehicle state deviation; The vehicle state of the target vehicle at the current moment is adjusted based on the control instruction.
2. The method according to claim 1, characterized in that Also includes: Based on the vehicle state prediction network, the state of the target vehicle is predicted to obtain a predicted value of the vehicle state of the target vehicle at a second preset future time after the current time; The target parameters in the vehicle state prediction value at the current moment and the vehicle state prediction value at the second preset future moment are identified, and abnormal state warning information is generated when the value of the target parameter is greater than the preset threshold value corresponding to the target parameter.
3. The method according to claim 2, characterized in that The method of performing state prediction on the target vehicle based on the vehicle state prediction network to obtain a predicted vehicle state of the target vehicle at a second preset future moment after the current moment includes: Taking the first preset historical moment vehicle state observations as an input sequence, inputting the input sequence into the vehicle state prediction network, and obtaining the current moment vehicle state prediction of the target vehicle at the current moment; The vehicle state prediction value at the current moment is used to replace the vehicle state observation value at the earliest historical moment in the input sequence with the time stamp, to obtain an updated input sequence, and the updated input sequence is input into the vehicle state prediction network to obtain the vehicle state prediction value at the next future moment of the vehicle state observation value at the current moment of the target vehicle; According to the preset prediction step, the above steps are repeated until the required vehicle state prediction value at the second preset future moment is obtained, wherein the preset prediction step corresponds to the vehicle state prediction value at the second preset future moment.
4. The method according to claim 1, characterized in that: The current moment vehicle state observation quantity includes the current moment vehicle observation position coordinates, the current moment vehicle observation speed, and the current moment vehicle observation acceleration; the current moment vehicle state prediction quantity includes the current moment vehicle prediction position coordinates, the current moment vehicle prediction speed, and the current moment vehicle prediction acceleration; The determining of the vehicle state deviation between the vehicle state observation at the current moment and the vehicle state prediction at the current moment includes: Calculating the deviation between the observed position coordinates of the vehicle at the current moment and the predicted position coordinates of the vehicle at the current moment to obtain the vehicle position coordinate deviation; Calculating the deviation between the current observed vehicle speed and the current predicted vehicle speed to obtain a vehicle speed deviation; Calculating the deviation between the observed acceleration of the vehicle at the current moment and the predicted acceleration of the vehicle at the current moment to obtain the vehicle acceleration deviation; The vehicle position coordinate deviation, the vehicle speed deviation, and the vehicle acceleration deviation are determined as the vehicle state deviation.
5. The method according to claim 1, characterized in that The generating a control instruction corresponding to the vehicle state deviation comprises: Using a membership function to perform fuzzy processing on the vehicle state deviation to generate a corresponding input fuzzy set; Reasoning the input fuzzy set according to preset rules to generate an output fuzzy set corresponding to the input fuzzy set; The output fuzzy set is defuzzified to generate the corresponding control instruction.
6. The method according to claim 1, characterized in that The control instruction includes a throttle control instruction and a steering control instruction, and the adjusting the vehicle state of the target vehicle at the current moment based on the control instruction includes: The throttle system of the target vehicle is controlled based on the throttle control instruction so that the vehicle speed deviation of the target vehicle is less than a first preset threshold value and the vehicle acceleration deviation is less than a second preset threshold value; the vehicle speed deviation is the deviation between the vehicle observed speed at the current moment in the vehicle state observation quantity at the current moment and the vehicle predicted speed at the current moment in the vehicle state prediction quantity at the current moment, and the vehicle acceleration deviation is the deviation between the vehicle observed acceleration at the current moment in the vehicle state observation quantity at the current moment and the vehicle predicted acceleration at the current moment in the vehicle state prediction quantity at the current moment; Based on the steering control instruction, the steering system of the target vehicle is controlled so that the vehicle position coordinate deviation of the target vehicle is less than a third preset threshold; the vehicle position coordinate deviation is the deviation between the vehicle observed position coordinates at the current moment in the vehicle state observation quantity at the current moment and the vehicle predicted position coordinates at the current moment in the vehicle state prediction quantity at the current moment.
7. The method according to claim 1, characterized in that The step of inputting the first preset historical moment vehicle state observation quantity into a vehicle state prediction network, and performing state prediction on the target vehicle based on the vehicle state prediction network to obtain a current moment vehicle state prediction quantity of the target vehicle at the current moment includes: Inputting the input sequence into the input layer of the vehicle state prediction network in sequence, and inputting the output sequence of the input layer into the hidden layer of the vehicle state prediction network in sequence; the input sequence is the vehicle state observation at the first preset historical moment; Based on the connection weights and thresholds between each gating unit and cell unit in the hidden layer, the features of the output sequence of the input layer are abstracted into a new dimensional space to extract the spatiotemporal variation features of the vehicle state of the target vehicle, and the spatiotemporal variation features of the vehicle state of the target vehicle are linearly divided, and the result of the linear division calculated by the hidden layer is input into the fully connected layer, wherein each gating unit includes an input gate, a forget gate and an output gate; Performing a nonlinear transformation on the input value of the fully connected layer to obtain an output value of the fully connected layer of a preset dimension; The output value of the fully connected layer is input into the regression output layer to predict the vehicle state based on the output value of the fully connected layer, so as to obtain the predicted value of the vehicle state of the target vehicle at the current moment.
8. A vehicle state control device, characterized in that: include: An acquisition module is configured to acquire a vehicle state observation value of a target vehicle at a first preset historical moment before a current moment and a vehicle state observation value of the target vehicle at a current moment; A vehicle state prediction module is configured to input the first preset historical moment vehicle state observation quantity into a vehicle state prediction network, perform state prediction on the target vehicle based on the vehicle state prediction network, and obtain a current moment vehicle state prediction quantity of the target vehicle at the current moment; a deviation correction module, configured to determine a vehicle state deviation between the vehicle state observation at the current moment and the vehicle state prediction at the current moment, and generate a control instruction corresponding to the vehicle state deviation; The control module is configured to adjust the vehicle state of the target vehicle at a current moment based on the control instruction.
9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.
10. A readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.