Light truck torque adaptive control method and system

Abstract

The present application provides a kind of light truck torque adaptive control method and system, method includes: obtaining light truck total mass and predicted torque trajectory;Light truck total mass, engine speed, road slope, yaw rate, mass center side angle, road curvature, road adhesion coefficient and light truck speed are combined into light truck state vector;Strategy neural network is constructed, to obtain front axle initial torque instruction and rear axle initial torque instruction in combination with predicted torque trajectory and light truck state vector;Front axle initial torque instruction and rear axle initial torque instruction are updated into front axle final torque instruction and rear axle final torque instruction based on control barrier function, to realize vehicle control.The light truck state vector is obtained by diversified data acquisition, combined with strategy neural network, based on the complex light truck state vector output, the torque distribution scheme that can simultaneously consider economy, power and stability is realized, and torque adaptive control is realized.

Description

Technical Field

[0001] This invention relates to the field of vehicle control technology, and in particular to a method and system for adaptive torque control of light trucks. Background Technology

[0002] Light trucks refer to light-duty trucks, which are currently widely used in urban delivery, short-distance transportation between urban and rural areas, and their operating conditions are characterized by frequent changes in load and diverse road conditions.

[0003] Specifically, frequent load changes refer to light trucks frequently switching between empty, half-loaded, and fully loaded states, with the total vehicle mass changing by more than 50% of the rated total mass, directly affecting power response, braking distance, and energy consumption. Diverse road conditions refer to the coexistence of urban roads, suburban highways, mountain slopes, curves, etc., with large variations in gradient, curvature, and coefficient of friction.

[0004] As the main control center of a vehicle, the core task of the vehicle controller is to optimize the distribution of drive torque between the front and rear axles in order to achieve multiple goals such as energy economy, driving power and driving stability. However, for light trucks, the operating conditions are complex and these goals are often mutually restrictive, making front and rear axle torque optimization a complex technical challenge. Summary of the Invention

[0005] To address the shortcomings of existing technologies, the present invention aims to provide a torque adaptive control method and system for light trucks, which solves the technical problem that, for light trucks, the operating conditions are complex and it is difficult to simultaneously achieve multiple objectives such as energy economy, driving power and driving stability by optimizing the distribution of drive torque between the front and rear axles.

[0006] To achieve the above objectives, in a first aspect, embodiments of this application provide a light truck torque adaptive control method, comprising the following steps:

[0007] The vertical acceleration signal of the wheel is obtained by the hub MEMS sensor, the total mass of the light truck is obtained based on the vertical acceleration signal, and the predicted torque trajectory is obtained based on the high-precision map.

[0008] The engine speed, road slope, yaw rate, center of gravity sideslip angle, road curvature, road surface adhesion coefficient, and light truck speed are obtained by the motor controller, inertial measurement unit, forward vision sensor, and wheel speed sensor, respectively. The total mass of the light truck, the engine speed, the road slope, the yaw rate, the center of gravity sideslip angle, the road curvature, the road surface adhesion coefficient, and the light truck speed are combined into a light truck state vector.

[0009] A strategy neural network is constructed to obtain the initial torque command of the front axle and the initial torque command of the rear axle through the predicted torque trajectory, the light truck state vector, and the strategy neural network.

[0010] A control barrier function is constructed based on the light truck state vector, the initial torque command of the front axle, and the torque command of the rear axle. The final torque command of the front axle and the final torque command of the rear axle are obtained based on the control barrier function. Vehicle control is performed based on the final torque command of the front axle and the final torque command of the rear axle.

[0011] Furthermore, the step of obtaining the total mass of the light truck based on the vertical acceleration signal includes:

[0012] The vertical acceleration signal is filtered to obtain a time-domain signal, and the time-domain signal is subjected to a Fourier transform to obtain the main frequency characteristics.

[0013] The vertical load of the wheel is obtained through the main frequency characteristic, and the total mass of the light truck is obtained through the vertical load.

[0014] Furthermore, the formula for obtaining the vertical load is:

[0015] ,

[0016] in, This represents the vertical load on the i-th wheel. This represents the dominant frequency characteristic of the i-th wheel. Indicates half-power bandwidth. , , All represent calibration parameters;

[0017] The formula for obtaining the total mass of the light truck is:

[0018] ,

[0019] in, Indicates the total mass of the light truck. Indicates the total number of wheels. It represents the acceleration due to gravity.

[0020] Furthermore, the step of obtaining the predicted torque trajectory based on a high-precision map includes:

[0021] Based on high-precision maps, predictive slope sequences and predictive curvature sequences are obtained. Slope constraints and curvature constraints are constructed using the predictive slope sequences and predictive curvature sequences. A torque prediction model including a longitudinal dynamics model and a motor energy consumption model is also constructed.

[0022] The objective function is to minimize the total equivalent energy consumption, and the torque prediction model is subject to vehicle speed constraints, slope constraints, and curvature constraints, so as to obtain the predicted torque trajectory through the torque prediction model.

[0023] Furthermore, the steps of acquiring engine speed, road gradient, yaw rate, center of gravity sideslip angle, road curvature, road surface adhesion coefficient, and light truck speed through the motor controller, inertial measurement unit, forward vision sensor, and wheel speed sensor respectively include:

[0024] The engine speed is obtained through the motor controller, and the longitudinal acceleration, yaw rate and center of gravity sideslip angle are obtained through the inertial measurement unit.

[0025] Wheel speed is obtained by wheel speed sensor, light truck speed is obtained by average of several wheel speeds, vehicle speed derivative is obtained based on light truck speed, and road slope is obtained by longitudinal acceleration and vehicle speed derivative.

[0026] Lane lines are acquired using a forward-looking vision sensor, and road curvature is obtained based on the lane lines;

[0027] The slip ratio is obtained based on the wheel speed of the drive wheel and the speed of the light truck. The longitudinal force of the tire is obtained through the hub MEMS sensor. The adhesion coefficient is calculated through the longitudinal force of the tire and the vertical load. A prediction curve is constructed based on the slip ratio and the calculated adhesion coefficient. The peak value in the prediction curve is selected as the road adhesion coefficient.

[0028] Furthermore, the step of obtaining the initial torque command for the front axle and the initial torque command for the rear axle through the predicted torque trajectory, the light truck state vector, and the strategy neural network includes:

[0029] The total mass of the light truck, the road gradient, the road curvature, and the road surface adhesion coefficient are used as constraints for the strategy neural network, and the engine speed, the yaw rate, the center of gravity offset angle, and the light truck speed are used as inputs for the strategy neural network to obtain the front axle predicted torque command and the rear axle predicted torque command.

[0030] A light truck dynamics simulation environment is constructed, with the predicted torque command of the front axle and the predicted torque command of the rear axle as the input to the light truck dynamics simulation environment to obtain an updated state vector. A reward function is constructed based on the light truck state vector and the updated state vector to update the setting parameters in the policy neural network through the reward function.

[0031] The light truck state vector is used as the input to the updated strategy neural network to obtain the initial torque command for the front axle and the initial torque command for the rear axle.

[0032] Furthermore, the reward function is:

[0033] ,

[0034] in, Represents the reward function, , , , These represent the weighting coefficients, This represents the instantaneous power of the front axle obtained based on the updated state vector. This represents the instantaneous power of the rear axle obtained based on the updated state vector. Indicates the reference power; This represents updating the light truck speed in the state vector. This represents the speed of the light truck in the light truck's state vector. This represents updating the yaw rate in the state vector. This represents the expected value of the yaw rate. Indicates the maximum yaw rate deviation. This represents the centroid offset angle in the updated state vector. Indicates the maximum centroid deflection angle. This indicates the predicted torque command for the front axle. This indicates the predicted torque command for the rear axle. This indicates the predicted torque trajectory.

[0035] Furthermore, the step of obtaining the final torque command of the front axle and the final torque command of the rear axle based on the control barrier function includes:

[0036] The control barrier function is linearly transformed to obtain a linear constraint function;

[0037] A quadratic programming function is constructed based on the initial torque of the front axle and the initial torque of the rear axle. The final torque command of the front axle and the final torque command of the rear axle are obtained based on the linear constraint function and the quadratic programming function.

[0038] Furthermore, the expression for the linear constraint function is:

[0039] ,

[0040] in, Represents the constraint matrix. Indicates the final torque command for the front axle. Indicates the final torque command for the rear axle. Indicates the transpose operator. Represents the target vector. Represents matrix multiplication;

[0041] The expression for the quadratic programming function is:

[0042] ,

[0043] in, This indicates the initial torque command for the front axle. This indicates the initial torque command for the rear axle. Represents the weight matrix. Indicates the transpose operator. Represents matrix multiplication. This represents the function to be minimized.

[0044] Secondly, embodiments of this application provide a light truck torque adaptive control system, applied to the light truck torque adaptive control method described in the first aspect above, the system comprising:

[0045] The first acquisition module is used to acquire the vertical acceleration signal of the wheel through the hub MEMS sensor, acquire the total mass of the light truck based on the vertical acceleration signal, and acquire the predicted torque trajectory based on the high-precision map.

[0046] The second acquisition module is used to acquire engine speed, road slope, yaw rate, center of gravity sideslip angle, road curvature, road surface adhesion coefficient and light truck speed through motor controller, inertial measurement unit, forward vision sensor and wheel speed sensor respectively, and combine the total mass of light truck, engine speed, road slope, yaw rate, center of gravity sideslip angle, road curvature, road surface adhesion coefficient and light truck speed into light truck state vector;

[0047] The third acquisition module is used to construct a strategy neural network to acquire the initial torque command of the front axle and the initial torque command of the rear axle through the predicted torque trajectory, the light truck state vector and the strategy neural network.

[0048] The execution module is used to construct a control barrier function based on the light truck state vector, the initial torque command of the front axle and the torque command of the rear axle, obtain the final torque command of the front axle and the final torque command of the rear axle based on the control barrier function, and perform vehicle control based on the final torque command of the front axle and the final torque command of the rear axle.

[0049] Thirdly, embodiments of this application provide a computer, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the light truck torque adaptive control method as described in the first aspect above.

[0050] Fourthly, embodiments of this application provide a storage medium storing a computer program thereon, which, when executed by a processor, implements the light truck torque adaptive control method as described in the first aspect above.

[0051] Compared with existing technologies, the beneficial effects of this invention are as follows: By acquiring the light truck state vector, diverse data collection is performed to address the complexity of operating conditions, comprehensively, realistically, and dynamically reflecting the state of the light truck, providing a data foundation for subsequent torque control; by constructing the strategy neural network, through its nonlinear fitting and self-learning capabilities, a torque distribution scheme that simultaneously considers economy, power, and stability is output from the complex light truck state vector, completing preliminary torque adaptive control; by constructing the control barrier function, safety optimization is performed on torque commands that do not meet the vehicle dynamics stability boundary, balancing the safety and reliability of torque control. Attached Figure Description

[0052] Figure 1 This is a flowchart of the light truck torque adaptive control method in the first embodiment of the present invention;

[0053] Figure 2 This is a structural block diagram of the light truck torque adaptive control system in the second embodiment of the present invention;

[0054] The following detailed description, in conjunction with the accompanying drawings, will further illustrate the present invention. Detailed Implementation

[0055] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Several embodiments of the invention are illustrated in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.

[0056] It should be noted that when a component is said to be "fixed to" another component, it can be directly on the other component or there may be an intervening component. When a component is said to be "connected to" another component, it can be directly connected to the other component or there may be an intervening component. The terms "vertical," "horizontal," "left," "right," and similar expressions used in this document are for illustrative purposes only.

[0057] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0058] Please see Figure 1 The first embodiment of the present invention provides a light truck torque adaptive control method, which includes the following steps:

[0059] S10: Obtain the vertical acceleration signal of the wheel through the hub MEMS sensor, obtain the total mass of the light truck based on the vertical acceleration signal, and obtain the predicted torque trajectory based on the high-precision map;

[0060] Step S10 includes:

[0061] S110: Filter the vertical acceleration signal to obtain a time-domain signal, and perform a Fourier transform on the time-domain signal to obtain the main frequency characteristics;

[0062] The filtering process refers to removing high-frequency noise and low-frequency drift from the vertical acceleration signal through bandpass filtering to form the time-domain signal. After performing a Fourier transform on the time-domain signal, its power spectrum can be obtained, and the main frequency characteristics can be obtained from the power spectral density (PSD).

[0063] S120: Obtain the vertical load of the wheel through the main frequency characteristics, and obtain the total mass of the light truck through the vertical load;

[0064] The formula for obtaining the vertical load is:

[0065] ,

[0066] in, This represents the vertical load on the i-th wheel. This represents the dominant frequency characteristic of the i-th wheel. Indicates half-power bandwidth. , , All of these represent calibration parameters; it should be noted that the shaft load-frequency response mapping model can be obtained in advance through bench testing to determine the values ​​of the calibration parameters.

[0067] The formula for obtaining the total mass of the light truck is:

[0068] ,

[0069] in, Indicates the total mass of the light truck. Indicates the total number of wheels. It represents the acceleration due to gravity.

[0070] S130: Obtain the predicted slope sequence and predicted curvature sequence based on the high-precision map, construct slope constraints and curvature constraints through the predicted slope sequence and predicted curvature sequence, and construct a torque prediction model including a longitudinal dynamics model and a motor energy consumption model.

[0071] The predicted slope sequence refers to a sequence of road slope values ​​corresponding to each point in the future driving trajectory obtained from a high-precision map. This sequence indicates whether the light truck is traveling uphill, downhill, or on a flat road in the future driving trajectory. The predicted curvature sequence refers to a sequence of road curvature corresponding to each point in the future driving trajectory obtained from a high-precision map. This sequence indicates whether the light truck is traveling straight, on a gentle curve, or on a sharp curve in the future driving trajectory. Based on the predicted slope and curvature sequences, the driving resistance can be obtained, thereby completing the construction of the longitudinal dynamics model. The motor energy consumption model describes the relationship between motor torque and energy consumption rate, and can typically be expressed through a lookup table or function.

[0072] S140: Taking the minimization of total equivalent energy consumption as the objective function, and applying vehicle speed constraints, slope constraints, and curvature constraints to the torque prediction model, so as to obtain the predicted torque trajectory through the torque prediction model;

[0073] The objective function is used to ensure that the total energy consumption in the future driving trajectory is minimized by the torque output. By applying the vehicle speed constraint, the slope constraint, and the curvature constraint, physical rules are imposed on the torque output to ensure that the output result is within a safe and reasonable range.

[0074] S20: The engine speed, road slope, yaw rate, center of gravity sideslip angle, road curvature, road surface adhesion coefficient, and light truck speed are obtained by the motor controller, inertial measurement unit, forward vision sensor, and wheel speed sensor, respectively. The total mass of the light truck, the engine speed, the road slope, the yaw rate, the center of gravity sideslip angle, the road curvature, the road surface adhesion coefficient, and the light truck speed are combined into a light truck state vector.

[0075] Specifically, step S20 includes:

[0076] S210: The engine speed is obtained through the motor controller, and the longitudinal acceleration, yaw rate and center of gravity sideslip angle are obtained through the inertial measurement unit;

[0077] The motor controller monitors the position and speed of the motor rotor in real time through a resolver or encoder to obtain the engine speed. Understandably, when the inertial measurement unit acquires the vertical acceleration signal through the accelerometer chip, it can simultaneously acquire the lateral acceleration and the longitudinal acceleration. Furthermore, the gyroscope chip in the inertial measurement unit can directly measure the angular velocity of the vehicle around the vertical rotation axis, i.e., the yaw rate. Further, after acquiring the yaw rate and the lateral acceleration, an observer can compare the difference between the vehicle motion predicted by the model and the actual motion based on the vehicle dynamics model, and use a Kalman filter algorithm to inversely calculate the sideslip angle of the center of mass that generated the motion.

[0078] S220: The wheel speed is obtained by a wheel speed sensor, the light truck speed is obtained by the average of several wheel speeds, the vehicle speed derivative is obtained based on the light truck speed, and the road slope is obtained by the longitudinal acceleration and the vehicle speed derivative.

[0079] The wheel speed sensor can be used to obtain the angular velocity of the wheel, and then combined with the effective radius of the wheel to obtain the wheel speed. The differential of the vehicle speed refers to the change in wheel speed within a short time interval, which can be obtained by dividing the difference in wheel speed between two samples by the product of the time difference between the two samples. Based on the wheel speeds of different wheels, the longitudinal acceleration of the light truck's center of gravity is obtained. The longitudinal acceleration of the light truck's center of gravity is compared with the longitudinal acceleration obtained by the inertial measurement unit to obtain the road slope.

[0080] S230: Obtain lane lines using a forward vision sensor, and obtain road curvature based on the lane lines;

[0081] In this embodiment, the forward vision sensor is a camera. The lane lines can be acquired through the forward vision sensor. It can be understood that the curvature of any point in the plane curve, i.e., the road curvature, can be calculated using the curvature formula.

[0082] S240: Obtain the slip ratio based on the wheel speed of the drive wheel and the speed of the light truck, obtain the longitudinal force of the tire through the wheel hub MEMS sensor, and obtain the calculated adhesion coefficient through the longitudinal force of the tire and the vertical load. Construct a prediction curve based on the slip ratio and the calculated adhesion coefficient, and select the peak value in the prediction curve as the road adhesion coefficient.

[0083] Specifically, slip ratio = (wheel speed of drive wheel - speed of light truck) / max(wheel speed of drive wheel, speed of light truck). Further, the calculated adhesion coefficient can be obtained by dividing the product of the longitudinal force of the tire and the vertical load. After obtaining the slip ratio and the calculated adhesion coefficient, the prediction curve can be constructed using the magic formula.

[0084] S30: Construct a strategy neural network to obtain the initial torque command of the front axle and the initial torque command of the rear axle through the predicted torque trajectory, the light truck state vector and the strategy neural network;

[0085] Specifically, step 30 includes:

[0086] S310: The total mass of the light truck, the road gradient, the road curvature, and the road surface adhesion coefficient are used as constraints of the strategy neural network, and the engine speed, the yaw rate, the center of gravity offset angle, and the light truck speed are used as inputs of the strategy neural network to obtain the front axle predicted torque command and the rear axle predicted torque command through the strategy neural network.

[0087] Understandably, the strategy neural network is used to output torque commands based on the light truck's state vector. Initially, the torque output is inaccurate, therefore, the strategy neural network needs to be optimized accordingly. The total mass of the light truck determines the energy consumption benchmark. Training the strategy neural network selects a relatively large torque that balances power and energy consumption. The road gradient and road curvature define the boundaries between safety and efficiency. The road gradient trains the strategy neural network to increase or decrease drive torque in advance when going uphill or downhill to maintain stable vehicle speed. The road curvature trains the strategy neural network to adjust torque distribution smoothly and in advance before entering a curve to ensure the vehicle curves in a safe and stable manner. The road surface adhesion coefficient defines the performance limit, training the strategy neural network to meet the performance limits under different road surface conditions, ensuring that the torque commands output by the strategy neural network meet the constraint requirements.

[0088] S320: Construct a light truck dynamics simulation environment, using the front axle predicted torque command and the rear axle predicted torque command as inputs to the light truck dynamics simulation environment to obtain an updated state vector, and construct a reward function based on the light truck state vector and the updated state vector to update the setting parameters in the policy neural network through the reward function;

[0089] The updated state vector is the same as the light truck state vector, only the time point is different, which will not be described again here. The reward function is:

[0090] ,

[0091] in, Represents the reward function, , , , These represent the weighting coefficients, This represents the instantaneous power of the front axle obtained based on the updated state vector. This represents the instantaneous power of the rear axle obtained based on the updated state vector. Indicates the reference power; This represents updating the light truck speed in the state vector. This represents the speed of the light truck in the light truck's state vector. This represents updating the yaw rate in the state vector. This represents the expected value of the yaw rate. Indicates the maximum yaw rate deviation. This represents the centroid offset angle in the updated state vector. Indicates the maximum centroid deflection angle. This indicates the predicted torque command for the front axle. This indicates the predicted torque command for the rear axle. This represents the predicted torque trajectory. Understandably, the reward function includes an energy consumption reward based on power, a dynamic performance reward based on acceleration, a stability reward based on yaw rate and center of mass yaw angle, and a tracking reward based on torque, to ensure that the output of the policy neural network can balance multiple objective requirements such as dynamic performance and energy consumption.

[0092] S330: Using the light truck state vector as the input to the updated strategy neural network, the initial torque command for the front axle and the initial torque command for the rear axle are obtained.

[0093] S40: Construct a control barrier function based on the light truck state vector, the initial torque command of the front axle and the torque command of the rear axle; obtain the final torque command of the front axle and the final torque command of the rear axle based on the control barrier function; and perform vehicle control based on the final torque command of the front axle and the final torque command of the rear axle.

[0094] The expression for the control barrier function is h(S, T) ≥ 0, where S represents the light truck state vector, and T represents the torque command vector including the initial torque command of the front axle and the initial torque command of the rear axle. However, the control barrier function is a nonlinear function, which is difficult to optimize directly. Therefore, it is necessary to perform a first-order Taylor expansion on the current light truck state vector and torque command vector. Specifically, step S40 includes:

[0095] S410: Perform a linear transformation on the control barrier function to obtain a linear constraint function;

[0096] The expression for the linear constraint function is:

[0097] ,

[0098] in, Represents the constraint matrix. Indicates the final torque command for the front axle. Indicates the final torque command for the rear axle. Indicates the transpose operator. Represents the target vector. This represents matrix multiplication.

[0099] S420: Construct a quadratic programming function based on the initial torque of the front axle and the initial torque of the rear axle, and obtain the final torque command of the front axle and the final torque command of the rear axle based on the linear constraint function and the quadratic programming function;

[0100] The expression for the quadratic programming function is:

[0101] ,

[0102] in, This indicates the initial torque command for the front axle. This indicates the initial torque command for the rear axle. Represents the weight matrix. Indicates the transpose operator. Represents matrix multiplication. This represents the function to be minimized.

[0103] By acquiring the light truck's state vector, diverse data collection was conducted to address the complexity of the operating conditions, comprehensively, realistically, and dynamically reflecting the light truck's state and providing a data foundation for subsequent torque control. By constructing the strategy neural network, its nonlinear fitting and self-learning capabilities are utilized to output a torque distribution scheme that simultaneously considers economy, power, and stability within the complex light truck state vector, achieving preliminary torque adaptive control. By constructing the control barrier function, torque commands that do not meet the vehicle's dynamic stability boundary are optimized for safety, ensuring both the safety and reliability of torque control.

[0104] Please see Figure 2 The second embodiment of the present invention provides a light truck torque adaptive control system, which is applied to the light truck torque adaptive control method described in the above embodiments, and will not be repeated hereafter. As used below, the terms "module," "unit," "subunit," etc., can refer to a combination of software and / or hardware that performs a predetermined function. Although the apparatus described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0105] The system includes:

[0106] The first acquisition module 10 is used to acquire the vertical acceleration signal of the wheel through the hub MEMS sensor, acquire the total mass of the light truck based on the vertical acceleration signal, and acquire the predicted torque trajectory based on the high-precision map.

[0107] The first acquisition module 10 includes:

[0108] The first unit is used to filter the vertical acceleration signal to obtain a time-domain signal, and to perform a Fourier transform on the time-domain signal to obtain the main frequency characteristics.

[0109] The second unit is used to obtain the vertical load of the wheel through the main frequency characteristics, and to obtain the total mass of the light truck through the vertical load;

[0110] The third unit is used to obtain the predicted slope sequence and the predicted curvature sequence based on the high-precision map, construct slope constraints and curvature constraints through the predicted slope sequence and the predicted curvature sequence, and construct a torque prediction model including a longitudinal dynamics model and a motor energy consumption model.

[0111] The fourth unit is used to apply vehicle speed constraints, slope constraints, and curvature constraints to the torque prediction model with the objective function of minimizing total equivalent energy consumption, so as to obtain the predicted torque trajectory through the torque prediction model.

[0112] The second acquisition module 20 is used to acquire engine speed, road slope, yaw rate, center of gravity sideslip angle, road curvature, road surface adhesion coefficient and light truck speed through the motor controller, inertial measurement unit, forward vision sensor and wheel speed sensor respectively, and combine the total mass of light truck, engine speed, road slope, yaw rate, center of gravity sideslip angle, road curvature, road surface adhesion coefficient and light truck speed into a light truck state vector;

[0113] The second acquisition module 20 includes:

[0114] The fifth unit is used to obtain the engine speed through the motor controller and the longitudinal acceleration, yaw rate and center of gravity sideslip angle through the inertial measurement unit.

[0115] The sixth unit is used to obtain the wheel speed through a wheel speed sensor, obtain the light truck speed by the average of several wheel speeds, obtain the vehicle speed derivative based on the light truck speed, and obtain the road slope by the longitudinal acceleration and the vehicle speed derivative.

[0116] The seventh unit is used to acquire lane lines through a forward-looking vision sensor and to acquire road curvature based on the lane lines;

[0117] The eighth unit is used to obtain the slip ratio based on the wheel speed of the drive wheel and the speed of the light truck, obtain the longitudinal force of the tire through the wheel hub MEMS sensor, and obtain the calculated adhesion coefficient through the longitudinal force of the tire and the vertical load. Based on the slip ratio and the calculated adhesion coefficient, a prediction curve is constructed, and the peak value in the prediction curve is selected as the road adhesion coefficient.

[0118] The third acquisition module 30 is used to construct a strategy neural network to acquire the initial torque command of the front axle and the initial torque command of the rear axle through the predicted torque trajectory, the light truck state vector and the strategy neural network.

[0119] The third acquisition module 30 includes:

[0120] The ninth unit is used to use the total mass of the light truck, the road slope, the road curvature and the road surface adhesion coefficient as constraints of the strategy neural network, and the engine speed, the yaw rate, the center of gravity offset angle and the light truck speed as inputs of the strategy neural network, so as to obtain the front axle predicted torque command and the rear axle predicted torque command through the strategy neural network.

[0121] The tenth unit is used to construct a light truck dynamics simulation environment. The predicted torque command of the front axle and the predicted torque command of the rear axle are used as inputs to the light truck dynamics simulation environment to obtain an updated state vector. A reward function is constructed based on the light truck state vector and the updated state vector to update the setting parameters in the policy neural network through the reward function.

[0122] The eleventh unit is used to obtain the initial torque command of the front axle and the initial torque command of the rear axle by using the light truck state vector as the input of the updated strategy neural network.

[0123] Execution module 40 is used to construct a control barrier function based on the light truck state vector, the initial torque command of the front axle and the torque command of the rear axle, obtain the final torque command of the front axle and the final torque command of the rear axle based on the control barrier function, and perform vehicle control based on the final torque command of the front axle and the final torque command of the rear axle.

[0124] The execution module 40 includes:

[0125] The twelfth unit is used to perform a linear transformation on the control barrier function to obtain a linear constraint function;

[0126] The thirteenth unit is used to construct a quadratic programming function based on the initial torque of the front axle and the initial torque of the rear axle, and to obtain the final torque command of the front axle and the final torque command of the rear axle based on the linear constraint function and the quadratic programming function.

[0127] The present invention also provides a computer, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the light truck torque adaptive control method as described in the above technical solution.

[0128] The present invention also provides a storage medium storing a computer program thereon, which, when executed by a processor, implements the light truck torque adaptive control method as described in the above technical solution.

[0129] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0130] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A method for adaptive torque control of a light truck, characterized in that, Includes the following steps: The vertical acceleration signal of the wheel is obtained by the hub MEMS sensor, the total mass of the light truck is obtained based on the vertical acceleration signal, and the predicted torque trajectory is obtained based on the high-precision map. The engine speed, road slope, yaw rate, center of gravity sideslip angle, road curvature, road surface adhesion coefficient, and light truck speed are obtained by the motor controller, inertial measurement unit, forward vision sensor, and wheel speed sensor, respectively. The total mass of the light truck, the engine speed, the road slope, the yaw rate, the center of gravity sideslip angle, the road curvature, the road surface adhesion coefficient, and the light truck speed are combined into a light truck state vector. A strategy neural network is constructed to obtain the initial torque command of the front axle and the initial torque command of the rear axle through the predicted torque trajectory, the light truck state vector, and the strategy neural network. A control barrier function is constructed based on the light truck state vector, the initial torque command of the front axle, and the initial torque command of the rear axle. The final torque command of the front axle and the final torque command of the rear axle are obtained based on the control barrier function. Vehicle control is performed based on the final torque command of the front axle and the final torque command of the rear axle.

2. The light truck torque adaptive control method according to claim 1, characterized in that, The step of obtaining the total mass of the light truck based on the vertical acceleration signal includes: The vertical acceleration signal is filtered to obtain a time-domain signal, and the time-domain signal is subjected to a Fourier transform to obtain the main frequency characteristics. The vertical load of the wheel is obtained through the main frequency characteristic, and the total mass of the light truck is obtained through the vertical load.

3. The light truck torque adaptive control method according to claim 2, characterized in that, The formula for obtaining the vertical load is: ， in, This represents the vertical load on the i-th wheel. This represents the dominant frequency characteristic of the i-th wheel. Indicates half-power bandwidth. , , All represent calibration parameters; The formula for obtaining the total mass of the light truck is: ， in, Indicates the total mass of the light truck. Indicates the total number of wheels. It represents the acceleration due to gravity.

4. The light truck torque adaptive control method according to claim 1, characterized in that, The steps for obtaining the predicted torque trajectory based on a high-precision map include: Based on high-precision maps, predictive slope sequences and predictive curvature sequences are obtained. Slope constraints and curvature constraints are constructed using the predictive slope sequences and predictive curvature sequences. A torque prediction model including a longitudinal dynamics model and a motor energy consumption model is also constructed. The objective function is to minimize the total equivalent energy consumption, and the torque prediction model is subject to vehicle speed constraints, slope constraints, and curvature constraints, so as to obtain the predicted torque trajectory through the torque prediction model.

5. The light truck torque adaptive control method according to claim 2, characterized in that, The steps of acquiring engine speed, road gradient, yaw rate, center of gravity sideslip angle, road curvature, road surface adhesion coefficient, and light truck speed through the motor controller, inertial measurement unit, forward vision sensor, and wheel speed sensor respectively include: The engine speed is obtained through the motor controller, and the longitudinal acceleration, yaw rate and center of gravity sideslip angle are obtained through the inertial measurement unit. Wheel speed is obtained by wheel speed sensor, light truck speed is obtained by average of several wheel speeds, vehicle speed derivative is obtained based on light truck speed, and road slope is obtained by longitudinal acceleration and vehicle speed derivative. Lane lines are acquired using a forward-looking vision sensor, and road curvature is obtained based on the lane lines; The slip ratio is obtained based on the wheel speed of the drive wheel and the speed of the light truck. The longitudinal force of the tire is obtained through the hub MEMS sensor. The adhesion coefficient is calculated through the longitudinal force of the tire and the vertical load. A prediction curve is constructed based on the slip ratio and the calculated adhesion coefficient. The peak value in the prediction curve is selected as the road adhesion coefficient.

6. The light truck torque adaptive control method according to claim 1, characterized in that, The step of obtaining the initial torque command for the front axle and the initial torque command for the rear axle through the predicted torque trajectory, the light truck state vector, and the strategy neural network includes: The total mass of the light truck, the road gradient, the road curvature, and the road surface adhesion coefficient are used as constraints for the strategy neural network, and the engine speed, the yaw rate, the center of gravity offset angle, and the light truck speed are used as inputs for the strategy neural network to obtain the front axle predicted torque command and the rear axle predicted torque command. A light truck dynamics simulation environment is constructed, with the predicted torque command of the front axle and the predicted torque command of the rear axle as the input to the light truck dynamics simulation environment to obtain an updated state vector. A reward function is constructed based on the light truck state vector and the updated state vector to update the setting parameters in the policy neural network through the reward function. The light truck state vector is used as the input to the updated strategy neural network to obtain the initial torque command for the front axle and the initial torque command for the rear axle.

7. The light truck torque adaptive control method according to claim 6, characterized in that, The reward function is: ， in, Represents the reward function, , , , These represent the weighting coefficients, This represents the instantaneous power of the front axle obtained based on the updated state vector. This represents the instantaneous power of the rear axle obtained based on the updated state vector. Indicates the reference power; This represents updating the light truck speed in the state vector. This represents the speed of the light truck in the light truck's state vector. This represents updating the yaw rate in the state vector. This represents the expected value of the yaw rate. Indicates the maximum yaw rate deviation. This represents the centroid offset angle in the updated state vector. Indicates the maximum centroid deflection angle. This indicates the predicted torque command for the front axle. This indicates the predicted torque command for the rear axle. This indicates the predicted torque trajectory.

8. The light truck torque adaptive control method according to claim 1, characterized in that, The step of obtaining the final torque command of the front axle and the final torque command of the rear axle based on the control barrier function includes: The control barrier function is linearly transformed to obtain a linear constraint function; A quadratic programming function is constructed based on the initial torque of the front axle and the initial torque of the rear axle. The final torque command of the front axle and the final torque command of the rear axle are obtained based on the linear constraint function and the quadratic programming function.

9. The light truck torque adaptive control method according to claim 8, characterized in that, The expression for the linear constraint function is: ， in, Represents the constraint matrix. Indicates the final torque command for the front axle. Indicates the final torque command for the rear axle. Indicates the transpose operator. Represents the target vector. Represents matrix multiplication; The expression for the quadratic programming function is: ， in, This indicates the initial torque command for the front axle. This indicates the initial torque command for the rear axle. Represents the weight matrix. Indicates the transpose operator. Represents matrix multiplication. This represents the function to be minimized.

10. A light truck torque adaptive control system, applied to the light truck torque adaptive control method as described in any one of claims 1 to 9, characterized in that, The system includes: The first acquisition module is used to acquire the vertical acceleration signal of the wheel through the hub MEMS sensor, acquire the total mass of the light truck based on the vertical acceleration signal, and acquire the predicted torque trajectory based on the high-precision map. The second acquisition module is used to acquire engine speed, road slope, yaw rate, center of gravity sideslip angle, road curvature, road surface adhesion coefficient and light truck speed through motor controller, inertial measurement unit, forward vision sensor and wheel speed sensor respectively, and combine the total mass of light truck, engine speed, road slope, yaw rate, center of gravity sideslip angle, road curvature, road surface adhesion coefficient and light truck speed into light truck state vector; The third acquisition module is used to construct a strategy neural network to acquire the initial torque command of the front axle and the initial torque command of the rear axle through the predicted torque trajectory, the light truck state vector and the strategy neural network. The execution module is used to construct a control barrier function based on the light truck state vector, the initial torque command of the front axle and the initial torque command of the rear axle, obtain the final torque command of the front axle and the final torque command of the rear axle based on the control barrier function, and perform vehicle control based on the final torque command of the front axle and the final torque command of the rear axle.

Images