A dual three-phase motor steer-by-wire system for heavy-duty transport equipment and its dual-layer composite control method

Abstract

This invention discloses a dual three-phase motor steer-by-wire system for heavy-duty transport equipment and its dual-layer composite control method, comprising: a steering wheel actuator unit that transmits the steering wheel angle signal to the electronic control unit, receives the torque signal sent by the electronic control unit, generates steering wheel return torque, and provides the driver with corresponding road feel information; and an electronic control unit that, based on the target steering angle, the actual steering angle, and the state of redundant sub-motors, establishes a triple redundant orthogonal decoupling model and constructs an upper-level steering angle tracking controller and a lower-level torque synchronization controller; and combines a superbasic function neural network adaptive interval observer and an Actor-Critic learning mechanism to achieve disturbance compensation and adaptive parameter adjustment, generating drive signals for each redundant sub-motor to drive the steering actuator unit to achieve wheel steering. This invention can suppress periodic disturbances under fault conditions, achieve fast and stable steering angle tracking, and maintain torque synchronization of the three redundant motors, improving the robustness and safety of the steer-by-wire actuator.

Description

Technical Field

[0001] This invention belongs to the field of drive-by-wire chassis control for heavy-duty transport equipment, specifically relating to a drive-by-wire steering system for dual three-phase motors in heavy-duty transport equipment and its dual-layer composite control method. Background Technology

[0002] With the rapid development of intelligent connected vehicles, new energy vehicles, and heavy-duty transportation equipment, vehicle steer-by-wire chassis technology (including steer-by-wire, brake-by-wire, and brake-by-wire) has become a key foundation for the development of intelligent vehicles. Steer-by-wire (SbW) systems eliminate the mechanical connection between the steering wheel and the steering wheels, relying on sensors, electronic control units, and motor actuators to complete steering actions. This significantly improves vehicle handling performance and autonomous driving adaptability in terms of response speed, control precision, and redundancy safety.

[0003] Compared to ordinary passenger vehicles, heavy-duty transport equipment typically features greater overall vehicle weight, higher axle load, heavier wheel-end load, greater steering resistance torque, and more complex operating environments. Its steering system faces higher performance loads and reliability requirements in low-speed, heavy-load steering, complex road maneuvering, long-term continuous operation, and high-safety-level tasks. Especially in applications such as mining transportation, port traction, engineering operations, and special transportation, steering failure in heavy-duty transport equipment can easily lead to path deviation, abnormal tire wear, decreased yaw stability, and even serious safety accidents. Therefore, steer-by-wire systems for heavy-duty transport equipment not only need high-precision angle tracking capabilities but also stronger redundant drive capabilities, fault tolerance, and continuous stable output capabilities.

[0004] However, most existing steer-by-wire systems use three-phase permanent magnet synchronous motors as actuators. When these motors experience single-phase open circuits, parameter mismatches, or electromagnetic interference, the output torque decreases significantly, resulting in noticeable periodic torque pulsations. This affects the vehicle's cornering performance and driving stability. With the continuous improvement of vehicle automation levels, steering actuators must possess higher levels of redundancy and fault tolerance to ensure the vehicle remains safe and controllable under extreme operating conditions and fault conditions.

[0005] To improve the safety of steer-by-wire systems, dual three-phase steering motors (DTP-PMSM) are increasingly being used in electric drive systems and steer-by-wire actuators due to their two independent three-phase windings, higher redundancy, and better fault tolerance. When one winding fails, the other can still maintain basic output. However, the traditional dual three-phase motor structure still has the following technical problems:

[0006] (1) When a single-phase open circuit or asymmetrical fault occurs in the winding, the motor output will show periodic torque pulsation related to the electrical angular frequency, resulting in asynchronous torque of the actuator and fluctuation in steering response;

[0007] (2) Existing fault-tolerant strategies usually rely on model switching, control reconfiguration or complex fault diagnosis algorithms, resulting in complex system design and poor real-time performance;

[0008] (3) Traditional sliding mode control and PI control have fixed parameters and do not have the ability to adapt to complex disturbances and dynamic changes in faults, making it difficult to meet the requirements of high-precision steer-by-wire for robustness and response speed.

[0009] In practical applications of steer-by-wire chassis, the vehicle steering actuator not only needs to maintain precise angle tracking under normal operating conditions, but also must maintain safe and reliable redundant output under conditions such as road disturbances, component aging, and electrical faults. Therefore, how to further improve redundancy utilization, effectively suppress periodic torque pulsation caused by faults, and achieve high-precision multi-motor torque synchronization control based on the dual three-phase motor structure are key technical problems that urgently need to be solved for intelligent steer-by-wire systems. Summary of the Invention

[0010] To address the shortcomings of the existing technology, the present invention aims to provide a dual three-phase motor steer-by-wire system for heavy-duty transport equipment and its dual-layer composite control method. This solves the problems of periodic torque pulsation, torque asynchrony, and increased angle tracking error caused by parameter mismatch, open-circuit faults, and external interference during the operation of existing steer-by-wire systems with three redundant motors. The present invention can suppress periodic disturbances under fault conditions, achieve fast and stable angle tracking, and maintain torque synchronization of the three redundant motors, thereby improving the robustness and safety of the steer-by-wire actuator.

[0011] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0012] The present invention provides a dual three-phase motor steer-by-wire system for heavy-duty transport equipment, comprising: a steering wheel actuator, a steering actuator, and an electronic control unit;

[0013] The steering wheel actuator transmits the steering wheel angle signal to the electronic control unit and receives the torque signal sent by the electronic control unit to generate the steering wheel return torque to provide the driver with corresponding road feel information.

[0014] The electronic control unit establishes a triple redundant orthogonal decoupling model based on the target steering angle, the actual steering angle, and the state of the redundant sub-motors, and constructs an upper-level steering angle tracking controller and a lower-level torque synchronization controller. It also combines a superbasic function neural network adaptive interval observer and an Actor-Critic learning mechanism to achieve disturbance compensation and parameter adaptive adjustment, thereby generating drive signals for each redundant sub-motor to drive the steering execution unit to achieve wheel steering.

[0015] Furthermore, in the system, both the steering wheel actuator and the steering actuator are electrically connected to the electronic control unit;

[0016] The steering wheel actuator includes: a steering wheel, a steering shaft, a steering wheel angle sensor, and a road feel motor; the steering wheel is mechanically connected to the road feel motor via the steering shaft; the steering wheel angle sensor is located on the steering shaft, which collects the actual steering angle and rate of change of the steering wheel in real time and sends the collected data to the electronic control unit; the road feel motor is electrically connected to the electronic control unit, and applies a controllable reaction torque to the steering wheel according to the control signal of the electronic control unit to generate road feel feedback;

[0017] The steering actuator includes: a dual three-phase steering motor, a dual three-phase steering motor angle sensor, a gear rack, a rack position sensor, a steering tie rod, and steering wheels; both ends of the gear rack are mechanically connected to the steering wheels via the steering tie rod; the dual three-phase steering motor is electrically connected to the electronic control unit, and the dual three-phase steering motor is mechanically connected to the gear rack; the rack position sensor is located on the gear rack and is used to collect the actual displacement or position change data of the rack and send the data to the electronic control unit; the dual three-phase steering motor angle sensor is located on the dual three-phase steering motor and is used to collect the rotor angle and angular velocity data of the dual three-phase motor in real time and send the data to the electronic control unit.

[0018] Furthermore, the dual three-phase steering motor adopts a triple redundant orthogonal decoupled dual three-phase steering motor. The six-phase windings of the dual three-phase steering motor undergo orthogonal transformation, dividing the six-phase windings into three independent redundant sub-motors.

[0019] The present invention provides a dual-layer composite control method for a dual three-phase motor steer-by-wire system for heavy-duty transport equipment. Based on the above system, the steps are as follows:

[0020] 1) Establish a structural model of a double three-phase steering motor steer-by-wire system with triple redundancy and orthogonal decoupling;

[0021] 2) Construct an upper-level steering angle tracking controller based on the error between the vehicle's desired steering angle and its actual steering angle;

[0022] 3) Design an adaptive interval observer based on a superbasic function neural network to estimate the upper and lower bounds of the system's comprehensive disturbance to generate a disturbance compensation amount. Input the disturbance compensation amount to the upper-level corner tracking controller to offset the influence of the comprehensive disturbance and obtain the compensated total control output.

[0023] 4) Construct a lower-level torque synchronization controller based on the torque tracking error of each redundant sub-motor, and adaptively adjust the parameters of the lower-level torque synchronization controller based on the Actor-Critic learning network; generate control signals for each redundant sub-motor based on the adjusted lower-level torque synchronization controller to achieve torque synchronization of each redundant sub-motor, and work with the upper-level steering angle tracking controller to complete the steer-by-wire control.

[0024] Further, in step 1), a two-degree-of-freedom vehicle lateral dynamics model is established with the vehicle sideslip angle and yaw rate as state variables to describe the relationship between the front wheel steering angle and the vehicle's lateral motion; a triple-redundant orthogonal decoupling model of the dual three-phase steering motor is established, and the six-phase windings of the dual three-phase steering motor are orthogonally transformed to divide the six-phase windings into three independent redundant sub-motors, forming three independent redundant systems that can output electromagnetic torque independently or collaboratively; the three redundant sub-motors, gear racks, front wheel loads, and external disturbances are modeled in a unified manner to obtain the structural model of the dual three-phase steering motor steer-by-wire system.

[0025] Further, step 1) specifically includes:

[0026] 11) Establish a two-degree-of-freedom vehicle lateral dynamics model, as follows:

[0027] ;

[0028] In the formula, For vehicle quality; The longitudinal speed of the vehicle; This refers to the vehicle's sideslip angle; The vehicle's yaw rate; Let be the moment of inertia about the z-axis; , These are the distances from the vehicle's center of gravity to the front and rear axles, respectively. , These are the lateral stiffness of the front and rear wheels, respectively.

[0029] 12) Establish a triple redundant orthogonal decoupling model for the dual three-phase steering motors, as follows:

[0030] By performing orthogonal transformation on the six-phase windings of the dual three-phase steering motor, the motor structure is decomposed into three sets of redundant sub-motors, forming three independent redundant systems: a redundant system composed of DB windings, a redundant system composed of EC windings, and a redundant system composed of FA windings. Here, A, B, C, D, E, and F represent the six stator phase windings of the dual three-phase steering motor. The magnetic field phases of the three sets of redundant sub-motors are spatially orthogonal and can output execution torque individually or collaboratively.

[0031] The current of each group of redundant sub-motors in the rotating coordinate system is expressed as:

[0032] ;

[0033] ;

[0034] ;

[0035] In the formula, , , These are the position angles of the d-axis, e-axis, and f-axis relative to the fundamental flux linkage, respectively. , These are the d-phase current and b-phase current of the redundant sub-motor corresponding to DB, respectively; , These are the e-phase and c-phase currents of the redundant sub-motor corresponding to EC; , Let f-phase current and a-phase current be the redundant sub-motor corresponding to FA;

[0036] The electromagnetic torque expression for each group of redundant sub-motors is:

[0037] ;

[0038] In the formula, For extreme logarithms, For the permanent magnet flux linkage of each group of redundant sub-motors;

[0039] Total torque of dual three-phase steering motors for:

[0040] ;

[0041] In the formula, , , The electromagnetic torque of the three sets of redundant sub-motors;

[0042] 13) Establish the structural model of the dual three-phase steering motor steer-by-wire system, as follows:

[0043] The dynamic equations of the dual three-phase steering motor steer-by-wire system are constructed as follows:

[0044] ;

[0045] ;

[0046] ;

[0047] In the formula, and Let be the d-axis and q-axis currents of the i-th redundant sub-motor, respectively. and They are and The differential; and The first The mechanical angular velocity of the redundant sub-motors and the dual three-phase steering motors. ; and The first The mechanical angle between the redundant sub-motors and the dual three-phase steering motors; This is the transmission ratio coefficient; For the first Stator inductance of redundant sub-motors; and These are the stator voltages along the d-axis and q-axis; Stator resistance; The rotor's electric angular velocity; and The equivalent moment of inertia and damping of a dual three-phase motor drive-by-wire steering system; The steering angle of the front wheels; This is the load torque;

[0048] Based on the dynamic equations of the above dual three-phase steering motor steer-by-wire system, the structural model of the dual three-phase steering motor steer-by-wire system is further constructed as follows:

[0049] ;

[0050] In the formula, Coulomb friction; This is the restoring torque; This refers to unmodeled disturbances and external interference. Torque pulsation caused by parameter mismatch and open circuit fault.

[0051] Further, in step 2), the desired vehicle steering angle input from the outside is received, and the actual vehicle steering angle is collected. The difference between the desired vehicle steering angle and the actual steering angle is calculated to obtain the steering angle error. The steering angle error is differentiated to obtain the rate of change of the steering angle error. The steering angle error and its rate of change are input into the error dynamic equation of the dual three-phase motor steer-by-wire system. A non-singular fast terminal sliding surface is designed, and an upper-level steering angle tracking controller is constructed based on the sliding surface. By adjusting the parameters of the upper-level steering angle tracking controller, the steering angle error and its rate of change are converged to 0, and the compensation current under disturbance is output.

[0052] Further, step 2) specifically includes:

[0053] The error between the vehicle's desired steering angle and its actual steering angle is expressed as:

[0054] ;

[0055] In the formula, The desired steering angle of the vehicle; This refers to the vehicle's actual steering angle. These are the observed values ​​of the steering angle; The observed value is the time derivative of the steering angle;

[0056] Based on the desired steering angle and the actual steering angle of the vehicle, the error dynamic equation of the dual three-phase motor steer-by-wire system can be obtained as follows:

[0057] ;

[0058] In the formula, For steering angle error, For steering angular velocity error, Non-zero constant coefficients; external disturbances , This represents the maximum value of the disturbance. For control input; Let be the expected value of the second time derivative of the steering angle;

[0059] The non-singular fast terminal sliding surface is designed as follows:

[0060] ;

[0061] In the formula, , and For the coefficients of the non-singular fast terminal sliding mode corner tracking controller; ; , It is a positive odd number. ;

[0062] The control law of the upper-level corner tracking controller is designed as follows:

[0063] ;

[0064] In the formula, ; For controller parameters; These are the coefficients for the upper-level corner tracking controller. ; It is a symbolic function.

[0065] Further, in step 3), the actual steering angle of the vehicle and the current output by the upper-level steering angle tracking controller are input to the adaptive interval observer based on the superbasic function neural network to calculate the estimated upper and lower bounds of the system's comprehensive disturbance, and generate a disturbance compensation amount according to the disturbance estimation interval; the disturbance compensation amount is input to the upper-level steering angle tracking controller to offset the influence of the comprehensive disturbance, and the compensated total control output is obtained.

[0066] Furthermore, step 3) specifically includes:

[0067] An adaptive interval observer based on a superbasic function neural network is designed to compensate for disturbances in the structural model of a dual three-phase steering motor steer-by-wire system, as detailed below:

[0068] The unknown disturbance in the dual three-phase motor steer-by-wire system is approximated using a superbasic function neural network, and the upper and lower bound estimates of the system disturbance are constructed as follows:

[0069] ;

[0070] In the formula, , and These represent the adaptive weight matrix, the upper bound of the adaptive weight matrix, and the lower bound of the adaptive weight matrix of the superbasic function neural network, respectively. , and Let represent the superbasic function, the upper bound of the superbasic function, and the lower bound of the superbasic function, respectively; , and This represents the approximation error, the upper bound of the approximation error, and the lower bound of the approximation error; This is the upper bound for estimating the overall system disturbance; This serves as the lower bound for estimating the overall system disturbance.

[0071] The state estimation dynamic equation of the adaptive interval observer based on the superbasic function neural network can be expressed as:

[0072] ;

[0073] in, , These represent the upper and lower bounds of the state, respectively; , The time derivatives of the upper and lower bounds of the state; For control input; The system's measurable output vector is represented by the gain matrix of the adaptive interval observer based on the hyperbasis function neural network, which is defined as follows: ,in , , and These are the state matrix, input matrix, output matrix, and observer gain matrix of the dynamic equation for the state estimation of an adaptive interval observer based on a superbasic function neural network. and These are the upper and lower bounds of the bounded approximation error generated when a superbasic function neural network approximates an unknown perturbation;

[0074] The upper and lower bound matrices of the weights in a hyperbasis function neural network are denoted as follows: and The update law is obtained from the following formula:

[0075] ;

[0076] In the formula, , These are the output matrices of the hidden layer of the hyperbasis function neural network. The upper and lower bounds; and Let be the upper bound weight gain matrix and the lower bound weight gain matrix of the superbasis function neural network weights, respectively, and satisfy . and The residual terms of the approximation error of the hyperbasis function neural network satisfy... and ; and This represents the upper and lower bound estimates of the output constructed by the adaptive interval observer based on a superbasic function neural network; and These are the weight decay coefficients in the upper bound weight update law and the lower bound weight update law, respectively. , .

[0077] Further, in step 4), the actual output torque of the three sets of redundant sub-motors is collected and subtracted from their respective reference torques to obtain the torque tracking error of each set of redundant sub-motors; the average value of the torque tracking error of the three sets of redundant sub-motors is used as the synchronization benchmark to construct the deviation of the torque tracking error of each set relative to the average value, which is used as the mean deviation coupling synchronization error; the torque compensation amount of each set of redundant sub-motors is generated based on the mean deviation coupling synchronization error; the torque tracking error and the mean deviation coupling synchronization error of each set of redundant sub-motors are used as state inputs to the Actor-Critic learning network to adjust the parameters of the lower-level torque synchronization controller online, so that the three sets of redundant sub-motors maintain torque output consistency under fault and disturbance conditions.

[0078] Further, step 4) specifically includes:

[0079] The torque tracking error of the i-th group of redundant sub-motors Defined as:

[0080] ;

[0081] In the formula, This is the reference torque for the i-th group of redundant sub-motors; This represents the actual output torque of the i-th redundant sub-motor.

[0082] The synchronization error of each set of redundant sub-motors is defined by the difference between its actual torque and that of the other two sets of redundant sub-motors, as follows:

[0083] ;

[0084] In the formula, and The actual torque of the two sets of redundant sub-motors other than redundant sub-motor i; The actual torque error of redundant sub-motor i and redundant sub-motor s. The actual torque error of redundant sub-motor i and redundant sub-motor r;

[0085] Torque compensation amount of the i-th group of redundant sub-motors Represented as:

[0086] ;

[0087] In the formula, , , These are the torque tracking errors of the three redundant sub-motors in the i, s, and r groups, respectively. and To compensate for the gain, the ratio of rotational inertia between each group of redundant sub-motors is determined. and The proportional and integral parameters of the PI controller; the average torque tracking error is defined as... ;

[0088] The difference between the torque tracking error of the i-th redundant sub-motor and the torque tracking errors of the other redundant sub-motors is as follows:

[0089] ;

[0090] In the formula, The difference in torque tracking error between redundant sub-motor i and redundant sub-motor s. This is the difference in torque tracking error between redundant sub-motor i and redundant sub-motor r;

[0091] The error between the torque tracking error of the i-th group of redundant sub-motors and the average torque tracking error of all redundant sub-motors for:

[0092] ;

[0093] PI parameters are dynamically adjusted based on real-time feedback; torque tracking error. Embedded state vector In this context, it is used for Actor-Critic network learning; the Actor-Critic network consists of: an Actor (policy network), a Critic (value network), and a stochastic action corrector; the Actor (policy network) maps the system state to candidate PI parameters of the i-th group of redundant sub-motors:

[0094] ;

[0095] in, Represents the candidate PI parameter vector. and These are the proportional parameter and integral parameter in the candidate PI parameter vector, respectively;

[0096] The random action corrector performs random perturbation correction based on the value function of the Critic (value network), and finally generates the following execution parameters:

[0097] ;

[0098] in, Indicates the first The actual execution PI parameter vector of the lower-level torque synchronization controller of the redundant sub-motor group; and These represent the proportional gain parameter and integral gain parameter in the actual PI parameter vector, respectively.

[0099] The beneficial effects of this invention are:

[0100] This invention orthogonally decouples the dual three-phase steering motors into three sets of two-phase redundant motors, and uses the three redundant motors to drive the steer-by-wire actuator together. Even under fault conditions such as open circuit in the windings, it can still maintain high output capacity and torque continuity, which significantly improves the redundancy, safety and fault tolerance of the steer-by-wire system.

[0101] This invention introduces an adaptive interval observer based on a superbasic function neural network and an upper-level angle tracking control structure, which can perform interval estimation and feedforward compensation for unmodeled dynamics, parameter uncertainties and external disturbances, so that the front wheel steering angle can still achieve fast and accurate tracking under disturbance and fault conditions, thereby improving the response speed and steady-state accuracy of the vehicle steering.

[0102] This invention employs an adaptive mean deviation coupled torque synchronization control strategy based on Actor-Critic reinforcement learning to adjust the PI gain parameter in the torque compensation of the three redundant motors online. This effectively suppresses torque asynchrony problems caused by parameter mismatch, open circuit faults, and load changes. While ensuring synchronization performance, it also takes into account dynamic response and robustness, further improving the stability and comfort of the steer-by-wire system under complex working conditions. Attached Figure Description

[0103] Figure 1 This is a schematic diagram of the structural principle of the system of the present invention.

[0104] Figure 2 This is a flowchart of the method of the present invention.

[0105] Figure 3 This is a schematic diagram of the corner tracking method of the present invention.

[0106] Figure 4 This is a schematic diagram of the corner tracking error in the method of the present invention. Detailed Implementation

[0107] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to embodiments and accompanying drawings. The content mentioned in the embodiments is not intended to limit the present invention.

[0108] Reference Figure 1 As shown, a dual three-phase motor steer-by-wire system for heavy-duty transport equipment according to the present invention includes: a steering wheel actuator, a steering actuator, and an electronic control unit;

[0109] The steering wheel actuator transmits the steering wheel angle signal to the electronic control unit and receives the torque signal sent by the electronic control unit to generate the steering wheel return torque to provide the driver with corresponding road feel information.

[0110] The electronic control unit establishes a triple redundant orthogonal decoupling model based on the target steering angle, the actual steering angle, and the state of the redundant sub-motors. It also constructs an upper-level steering angle tracking controller and a lower-level torque synchronization controller. Furthermore, it combines a superbasic function neural network adaptive interval observer and an Actor-Critic learning mechanism to achieve disturbance compensation and parameter adaptive adjustment, thereby generating drive signals for each redundant sub-motor to drive the steering execution unit to achieve wheel steering.

[0111] Specifically, in the system, both the steering wheel actuator and the steering actuator are electrically connected to the electronic control unit;

[0112] The steering wheel actuator 1 includes: a steering wheel 4, a steering shaft 5, a steering wheel angle sensor 6, and a road feel motor 7; the steering wheel 4 is mechanically connected to the road feel motor 7 via the steering shaft 5; the steering wheel angle sensor 6 is mounted on the steering shaft, which collects the actual steering angle and rate of change of the steering wheel in real time and sends the collected data to the electronic control unit 3; the road feel motor 7 is electrically connected to the electronic control unit 3, and applies a controllable reaction torque to the steering wheel 4 according to the control signal of the electronic control unit 3 to generate road feel feedback;

[0113] The steering actuator 2 includes: a dual three-phase steering motor 8, a dual three-phase steering motor angle sensor 9, a gear rack 10, a rack position sensor 11, a steering tie rod 12, and a steering wheel 13; both ends of the gear rack 10 are mechanically connected to the steering wheel 13 via the steering tie rod 12; the dual three-phase steering motor 8 is electrically connected to the electronic control unit 3, and the dual three-phase steering motor 8 is mechanically connected to the gear rack 10; the rack position sensor 11 is mounted on the gear rack and is used to collect the actual displacement or position change data of the rack and send the data to the electronic control unit 3; the dual three-phase steering motor angle sensor 9 is mounted on the dual three-phase steering motor 8 and is used to collect the rotor angle and angular velocity data of the dual three-phase motor in real time and send the data to the electronic control unit 3.

[0114] In the example, the dual three-phase steering motor 8 adopts a triple redundant orthogonal decoupled dual three-phase steering motor. The six-phase windings of the dual three-phase steering motor undergo orthogonal transformation, dividing the six-phase windings into three groups of independent redundant sub-motors.

[0115] Reference Figures 2 to 4 As shown, the present invention discloses a dual-layer composite control method for a dual three-phase motor drive-by-wire steering system for heavy-duty transport equipment. Based on the above system, the steps are as follows:

[0116] 1) Establish a structural model of a double three-phase steering motor steer-by-wire system with triple redundancy and orthogonal decoupling;

[0117] In step 1), a two-degree-of-freedom vehicle lateral dynamics model is established with the vehicle sideslip angle and yaw rate as state variables to describe the relationship between the front wheel steering angle and the vehicle's lateral motion. A triple-redundant orthogonal decoupling model of the dual three-phase steering motor is established. The six-phase windings of the dual three-phase steering motor are orthogonally transformed and divided into three independent redundant sub-motors, forming three independent redundant systems that can output electromagnetic torque independently or collaboratively. The three redundant sub-motors, gear racks, front wheel loads, and external disturbances are modeled in a unified manner to obtain the structural model of the dual three-phase steering motor steer-by-wire system.

[0118] It should be noted that step 1) specifically includes:

[0119] 11) Establish a two-degree-of-freedom vehicle lateral dynamics model, as follows:

[0120] ;

[0121] In the formula, For vehicle quality; The longitudinal speed of the vehicle; This refers to the vehicle's sideslip angle; The vehicle's yaw rate; Let be the moment of inertia about the z-axis; , These are the distances from the vehicle's center of gravity to the front and rear axles, respectively. , These are the lateral stiffness of the front and rear wheels, respectively.

[0122] 12) Establish a triple redundant orthogonal decoupling model for the dual three-phase steering motors, as follows:

[0123] By performing orthogonal transformation on the six-phase windings of the dual three-phase steering motor, the motor structure is decomposed into three sets of redundant sub-motors, forming three independent redundant systems: a redundant system composed of DB windings, a redundant system composed of EC windings, and a redundant system composed of FA windings. Here, A, B, C, D, E, and F represent the six stator phase windings of the dual three-phase steering motor. The magnetic field phases of the three sets of redundant sub-motors are spatially orthogonal and can output execution torque individually or collaboratively.

[0124] The current of each group of redundant sub-motors in the rotating coordinate system is expressed as:

[0125] ;

[0126] ;

[0127] ;

[0128] In the formula, , , These are the position angles of the d-axis, e-axis, and f-axis relative to the fundamental flux linkage, respectively. , These are the d-phase current and b-phase current of the redundant sub-motor corresponding to DB, respectively; , These are the e-phase and c-phase currents of the redundant sub-motor corresponding to EC; , Let f-phase current and a-phase current be the redundant sub-motor corresponding to FA;

[0129] The electromagnetic torque expression for each group of redundant sub-motors is:

[0130] ;

[0131] In the formula, For extreme logarithms, For the permanent magnet flux linkage of each group of redundant sub-motors;

[0132] Total torque of dual three-phase steering motors for:

[0133] ;

[0134] In the formula, , , The electromagnetic torque of the three sets of redundant sub-motors;

[0135] 13) Establish the structural model of the dual three-phase steering motor steer-by-wire system, as follows:

[0136] The dynamic equations of the dual three-phase steering motor steer-by-wire system are constructed as follows:

[0137] ;

[0138] ;

[0139] ;

[0140] In the formula, and Let be the d-axis and q-axis currents of the i-th redundant sub-motor, respectively. and They are and The differential; and The first The mechanical angular velocity of the redundant sub-motors and the dual three-phase steering motors. ; and The first The mechanical angle between the redundant sub-motors and the dual three-phase steering motors; This is the transmission ratio coefficient; For the first Stator inductance of redundant sub-motors; and These are the stator voltages along the d-axis and q-axis; Stator resistance; The rotor's electric angular velocity; and The equivalent moment of inertia and damping of a dual three-phase motor drive-by-wire steering system; The steering angle of the front wheels; This is the load torque;

[0141] Based on the dynamic equations of the above dual three-phase steering motor steer-by-wire system, the structural model of the dual three-phase steering motor steer-by-wire system is further constructed as follows:

[0142] ;

[0143] In the formula, Coulomb friction; This is the restoring torque; This refers to unmodeled disturbances and external interference. Torque pulsation caused by parameter mismatch and open circuit fault.

[0144] 2) Construct an upper-level steering angle tracking controller based on the error between the vehicle's desired steering angle and its actual steering angle;

[0145] In step 2), the desired vehicle steering angle is received from an external input, and the actual vehicle steering angle is collected. The difference between the desired and actual vehicle steering angles is calculated to obtain the steering angle error. The steering angle error is differentiated to obtain the rate of change of the steering angle error. The steering angle error and its rate of change are input into the error dynamic equation of the dual three-phase motor steer-by-wire system. A non-singular fast terminal sliding surface is designed, and an upper-level steering angle tracking controller is constructed based on the sliding surface. By adjusting the parameters of the upper-level steering angle tracking controller, the steering angle error and its rate of change converge to 0, and the compensation current under disturbance is output.

[0146] It should be noted that step 2) specifically includes:

[0147] The error between the vehicle's desired steering angle and its actual steering angle is expressed as:

[0148] ;

[0149] In the formula, The desired steering angle of the vehicle; This refers to the vehicle's actual steering angle. These are the observed values ​​of the steering angle; The observed value is the time derivative of the steering angle;

[0150] Based on the desired steering angle and the actual steering angle of the vehicle, the error dynamic equation of the dual three-phase motor steer-by-wire system can be obtained as follows:

[0151] ;

[0152] In the formula, For steering angle error, For steering angular velocity error, Non-zero constant coefficients; external disturbances , This represents the maximum value of the disturbance. For control input; Let be the expected value of the second time derivative of the steering angle;

[0153] The non-singular fast terminal sliding surface is designed as follows:

[0154] ;

[0155] In the formula, , and For the coefficients of the non-singular fast terminal sliding mode corner tracking controller; ; , It is a positive odd number. ;

[0156] The control law of the upper-level corner tracking controller is designed as follows:

[0157] ;

[0158] In the formula, ; For controller parameters; These are the coefficients for the upper-level corner tracking controller. ; It is a symbolic function.

[0159] 3) Design an adaptive interval observer based on a superbasic function neural network to estimate the upper and lower bounds of the system's comprehensive disturbance to generate a disturbance compensation amount. Input the disturbance compensation amount to the upper-level corner tracking controller to offset the influence of the comprehensive disturbance and obtain the compensated total control output.

[0160] In step 3), the actual steering angle of the vehicle and the current output by the upper-level steering angle tracking controller are input to the adaptive interval observer based on the superbasic function neural network to calculate the estimated upper and lower bounds of the system's comprehensive disturbance, and generate a disturbance compensation amount based on the disturbance estimation interval; the disturbance compensation amount is input to the upper-level steering angle tracking controller to offset the influence of the comprehensive disturbance, and the compensated total control output is obtained.

[0161] It should be noted that step 3) specifically includes:

[0162] An adaptive interval observer based on a superbasic function neural network is designed to compensate for disturbances in the structural model of a dual three-phase steering motor steer-by-wire system, as detailed below:

[0163] The unknown disturbance in the dual three-phase motor steer-by-wire system is approximated using a superbasic function neural network, and the upper and lower bound estimates of the system disturbance are constructed as follows:

[0164] ;

[0165] In the formula, , and These represent the adaptive weight matrix, the upper bound of the adaptive weight matrix, and the lower bound of the adaptive weight matrix of the superbasic function neural network, respectively. , and Let represent the superbasic function, the upper bound of the superbasic function, and the lower bound of the superbasic function, respectively; , and This represents the approximation error, the upper bound of the approximation error, and the lower bound of the approximation error; This is the upper bound for estimating the overall system disturbance; This serves as the lower bound for estimating the overall system disturbance.

[0166] The state estimation dynamic equation of the adaptive interval observer based on the superbasic function neural network can be expressed as:

[0167] ;

[0168] in, , These represent the upper and lower bounds of the state, respectively; , The time derivatives of the upper and lower bounds of the state; For control input; The system's measurable output vector is represented by the gain matrix of the adaptive interval observer based on the hyperbasis function neural network, which is defined as follows: ,in , , and These are the state matrix, input matrix, output matrix, and observer gain matrix of the dynamic equation for the state estimation of an adaptive interval observer based on a superbasic function neural network. and These are the upper and lower bounds of the bounded approximation error generated when a superbasic function neural network approximates an unknown perturbation;

[0169] The upper and lower bound matrices of the weights in a hyperbasis function neural network are denoted as follows: and The update law is obtained from the following formula:

[0170] ;

[0171] In the formula, , These are the output matrices of the hidden layer of the hyperbasis function neural network. The upper and lower bounds; and Let be the upper bound weight gain matrix and the lower bound weight gain matrix of the superbasis function neural network weights, respectively, and satisfy . and The residual terms of the approximation error of the hyperbasis function neural network satisfy... and ; and This represents the upper and lower bound estimates of the output constructed by the adaptive interval observer based on a superbasic function neural network; and These are the weight decay coefficients in the upper bound weight update law and the lower bound weight update law, respectively. , .

[0172] 4) Construct a lower-level torque synchronization controller based on the torque tracking error of each redundant sub-motor, and adaptively adjust the parameters of the lower-level torque synchronization controller based on the Actor-Critic learning network; generate control signals for each redundant sub-motor based on the adjusted lower-level torque synchronization controller to achieve torque synchronization of each redundant sub-motor, and work with the upper-level steering angle tracking controller to complete the steer-by-wire control.

[0173] In step 4), the actual output torque of the three sets of redundant sub-motors is collected and subtracted from their respective reference torques to obtain the torque tracking error of each set of redundant sub-motors. The average torque tracking error of the three sets of redundant sub-motors is used as the synchronization benchmark to construct the deviation of each set of torque tracking errors from the average value, which is used as the mean deviation coupling synchronization error. The torque compensation amount of each set of redundant sub-motors is generated based on the mean deviation coupling synchronization error. The torque tracking error and mean deviation coupling synchronization error of each set of redundant sub-motors are used as state inputs to the Actor-Critic learning network to adjust the parameters of the lower-level torque synchronization controller online, so that the three sets of redundant sub-motors maintain torque output consistency under fault and disturbance conditions.

[0174] It should be noted that step 4) specifically includes:

[0175] The torque tracking error of the i-th group of redundant sub-motors Defined as:

[0176] ;

[0177] In the formula, This is the reference torque for the i-th group of redundant sub-motors; This represents the actual output torque of the i-th redundant sub-motor.

[0178] The synchronization error of each set of redundant sub-motors is defined by the difference between its actual torque and that of the other two sets of redundant sub-motors, as follows:

[0179] ;

[0180] In the formula, and The actual torque of the two sets of redundant sub-motors other than redundant sub-motor i; The actual torque error of redundant sub-motor i and redundant sub-motor s. The actual torque error of redundant sub-motor i and redundant sub-motor r;

[0181] Torque compensation amount of the i-th group of redundant sub-motors Represented as:

[0182] ;

[0183] In the formula, , , These are the torque tracking errors of the three redundant sub-motors in the i, s, and r groups, respectively. and To compensate for the gain, the ratio of rotational inertia between each group of redundant sub-motors is determined. and The proportional and integral parameters of the PI controller; the average torque tracking error is defined as... ;

[0184] The difference between the torque tracking error of the i-th redundant sub-motor and the torque tracking errors of the other redundant sub-motors is as follows:

[0185] ;

[0186] In the formula, The difference in torque tracking error between redundant sub-motor i and redundant sub-motor s. This is the difference in torque tracking error between redundant sub-motor i and redundant sub-motor r;

[0187] The error between the torque tracking error of the i-th group of redundant sub-motors and the average torque tracking error of all redundant sub-motors for:

[0188] ;

[0189] PI parameters are dynamically adjusted based on real-time feedback; torque tracking error. Embedded state vector In this context, it is used for Actor-Critic network learning; the Actor-Critic network consists of: an Actor (policy network), a Critic (value network), and a stochastic action corrector; the Actor (policy network) maps the system state to candidate PI parameters of the i-th group of redundant sub-motors:

[0190] ;

[0191] in, Represents the candidate PI parameter vector. and These are the proportional parameter and integral parameter in the candidate PI parameter vector, respectively;

[0192] The random action corrector performs random perturbation correction based on the value function of the Critic (value network), and finally generates the following execution parameters:

[0193] ;

[0194] in, Indicates the first The actual execution PI parameter vector of the lower-level torque synchronization controller of the redundant sub-motor group; and These represent the proportional gain parameter and integral gain parameter in the actual PI parameter vector, respectively.

[0195] This invention has many specific applications. The above description is only a preferred embodiment of this invention. It should be noted that for those skilled in the art, several improvements can be made without departing from the principle of this invention, and these improvements should also be considered within the scope of protection of this invention.

Claims

1. A dual three-phase motor drive-by-wire steering system for heavy-duty transport equipment, characterized in that, include: Steering wheel actuator, steering actuator, and electronic control unit; The steering wheel actuator transmits the steering wheel angle signal to the electronic control unit and receives the torque signal sent by the electronic control unit to generate the steering wheel return torque to provide the driver with corresponding road feel information. The electronic control unit establishes a triple redundant orthogonal decoupling model based on the target steering angle, the actual steering angle, and the state of the redundant sub-motors, and constructs an upper-level steering angle tracking controller and a lower-level torque synchronization controller. It also combines a superbasic function neural network adaptive interval observer and an Actor-Critic learning mechanism to achieve disturbance compensation and parameter adaptive adjustment, thereby generating drive signals for each redundant sub-motor to drive the steering execution unit to achieve wheel steering.

2. The heavy-duty transport equipment dual three-phase motor drive-by-wire steering system according to claim 1, characterized in that, In the system, both the steering wheel actuator and the steering actuator are electrically connected to the electronic control unit. The steering wheel actuator includes: a steering wheel, a steering shaft, a steering wheel angle sensor, and a road feel motor; the steering wheel is mechanically connected to the road feel motor via the steering shaft; the steering wheel angle sensor is located on the steering shaft, which collects the actual steering angle and rate of change of the steering wheel in real time and sends the collected data to the electronic control unit; the road feel motor is electrically connected to the electronic control unit, and applies a controllable reaction torque to the steering wheel according to the control signal of the electronic control unit to generate road feel feedback; The steering actuator includes: a dual three-phase steering motor, a dual three-phase steering motor angle sensor, a gear rack, a rack position sensor, a steering tie rod, and steering wheels; both ends of the gear rack are mechanically connected to the steering wheels via the steering tie rod; the dual three-phase steering motor is electrically connected to the electronic control unit, and the dual three-phase steering motor is mechanically connected to the gear rack; the rack position sensor is located on the gear rack and is used to collect the actual displacement or position change data of the rack and send the data to the electronic control unit; the dual three-phase steering motor angle sensor is located on the dual three-phase steering motor and is used to collect the rotor angle and angular velocity data of the dual three-phase motor in real time and send the data to the electronic control unit.

3. A two-layer composite control method for a dual three-phase motor drive-by-wire steering system for heavy-duty transport equipment, based on the system described in any one of claims 1-2, characterized in that, The steps are as follows: 1) Establish a structural model of a double three-phase steering motor steer-by-wire system with triple redundancy and orthogonal decoupling; 2) Construct an upper-level steering angle tracking controller based on the error between the vehicle's desired steering angle and its actual steering angle; 3) Design an adaptive interval observer based on a superbasic function neural network to estimate the upper and lower bounds of the system's comprehensive disturbance to generate a disturbance compensation amount. Input the disturbance compensation amount to the upper-level corner tracking controller to offset the influence of the comprehensive disturbance and obtain the compensated total control output. 4) Construct a lower-level torque synchronization controller based on the torque tracking error of each redundant sub-motor, and adaptively adjust the parameters of the lower-level torque synchronization controller based on the Actor-Critic learning network; generate control signals for each redundant sub-motor based on the adjusted lower-level torque synchronization controller to achieve torque synchronization of each redundant sub-motor, and work with the upper-level steering angle tracking controller to complete the steer-by-wire control.

4. The dual-layer composite control method for the dual three-phase motor drive-by-wire steering system of heavy-duty transport equipment according to claim 3, characterized in that, In step 1), a two-degree-of-freedom vehicle lateral dynamics model is established with the vehicle sideslip angle and yaw rate as state variables to describe the relationship between the front wheel steering angle and the vehicle's lateral motion. A triple-redundant orthogonal decoupling model of the dual three-phase steering motor is established. The six-phase windings of the dual three-phase steering motor are orthogonally transformed and divided into three independent redundant sub-motors, forming three independent redundant systems that can output electromagnetic torque independently or collaboratively. The three redundant sub-motors, gear racks, front wheel loads, and external disturbances are modeled in a unified manner to obtain the structural model of the dual three-phase steering motor steer-by-wire system.

5. The dual-layer composite control method for the dual three-phase motor drive-by-wire steering system of heavy-duty transport equipment according to claim 4, characterized in that, Step 1) specifically includes: 11) Establish a two-degree-of-freedom vehicle lateral dynamics model, as follows: ； In the formula, For vehicle quality; The longitudinal speed of the vehicle; This refers to the vehicle's sideslip angle; The vehicle's yaw rate; Let be the moment of inertia about the z-axis; , These are the distances from the vehicle's center of gravity to the front and rear axles, respectively. , These are the lateral stiffness of the front and rear wheels, respectively. 12) Establish a triple redundant orthogonal decoupling model for the dual three-phase steering motors, as follows: By performing orthogonal transformation on the six-phase windings of the dual three-phase steering motor, the motor structure is decomposed into three sets of redundant sub-motors, forming three independent redundant systems: a redundant system composed of DB windings, a redundant system composed of EC windings, and a redundant system composed of FA windings. Here, A, B, C, D, E, and F represent the six stator phase windings of the dual three-phase steering motor. The magnetic field phases of the three sets of redundant sub-motors are spatially orthogonal and can output execution torque individually or collaboratively. The current of each group of redundant sub-motors in the rotating coordinate system is expressed as: ； ； ； In the formula, , , These are the position angles of the d-axis, e-axis, and f-axis relative to the fundamental flux linkage, respectively. , These are the d-phase current and b-phase current of the redundant sub-motor corresponding to DB, respectively; , These are the e-phase and c-phase currents of the redundant sub-motor corresponding to EC; , Let f-phase current and a-phase current be the redundant sub-motor corresponding to FA; The electromagnetic torque expression for each group of redundant sub-motors is: ； In the formula, For extreme logarithms, For the permanent magnet flux linkage of each group of redundant sub-motors; Total torque of dual three-phase steering motors for: ； In the formula, , , The electromagnetic torque of the three sets of redundant sub-motors; 13) Establish the structural model of the dual three-phase steering motor steer-by-wire system, as follows: The dynamic equations of the dual three-phase steering motor steer-by-wire system are constructed as follows: ； ； ； In the formula, and Let be the d-axis and q-axis currents of the i-th redundant sub-motor, respectively. and They are and The differential; and The first The mechanical angular velocity of the redundant sub-motors and the dual three-phase steering motors. ; and The first The mechanical angle between the redundant sub-motors and the dual three-phase steering motors; This is the transmission ratio coefficient; For the first Stator inductance of redundant sub-motors; and These are the stator voltages along the d-axis and q-axis; Stator resistance; The rotor's electric angular velocity; and The equivalent moment of inertia and damping of a dual three-phase motor drive-by-wire steering system; The steering angle of the front wheels; This is the load torque; Based on the dynamic equations of the above dual three-phase steering motor steer-by-wire system, the structural model of the dual three-phase steering motor steer-by-wire system is further constructed as follows: ； In the formula, Coulomb friction; This is the restoring torque; This refers to unmodeled disturbances and external interference. Torque pulsation caused by parameter mismatch and open circuit fault.

6. The dual-layer composite control method for the dual three-phase motor steer-by-wire system of heavy-duty transport equipment according to claim 5, characterized in that, In step 2), the desired vehicle steering angle is received from an external input, and the actual vehicle steering angle is collected. The difference between the desired and actual vehicle steering angles is calculated to obtain the steering angle error. The steering angle error is differentiated to obtain the rate of change of the steering angle error. The steering angle error and its rate of change are input into the error dynamic equation of the dual three-phase motor steer-by-wire system. A non-singular fast terminal sliding surface is designed, and an upper-level steering angle tracking controller is constructed based on the sliding surface. By adjusting the parameters of the upper-level steering angle tracking controller, the steering angle error and its rate of change converge to 0, and the compensation current under disturbance is output.

7. The dual-layer composite control method for the dual three-phase motor drive-by-wire steering system of heavy-duty transport equipment according to claim 6, characterized in that, Step 2) specifically includes: The error between the vehicle's desired steering angle and its actual steering angle is expressed as: ； In the formula, The desired steering angle of the vehicle; This refers to the vehicle's actual steering angle. These are the observed values ​​of the steering angle; The observed value is the time derivative of the steering angle; Based on the desired steering angle and the actual steering angle of the vehicle, the error dynamic equation of the dual three-phase motor steer-by-wire system can be obtained as follows: ； In the formula, For steering angle error, For steering angular velocity error, Non-zero constant coefficients; external disturbances , This represents the maximum value of the disturbance. For control input; Let be the expected value of the second time derivative of the steering angle; The non-singular fast terminal sliding surface is designed as follows: ； In the formula, , and For the coefficients of the non-singular fast terminal sliding mode corner tracking controller; ; , It is a positive odd number. ; The control law of the upper-level corner tracking controller is designed as follows: ； In the formula, ; For controller parameters; These are the coefficients for the upper-level corner tracking controller. ; It is a symbolic function.

8. The dual-layer composite control method for the dual three-phase motor drive-by-wire steering system of heavy-duty transport equipment according to claim 7, characterized in that, In step 3), the actual steering angle of the vehicle and the current output by the upper-level steering angle tracking controller are input to the adaptive interval observer based on the superbasic function neural network to calculate the estimated upper and lower bounds of the system's comprehensive disturbance, and generate a disturbance compensation amount based on the disturbance estimation interval; the disturbance compensation amount is input to the upper-level steering angle tracking controller to offset the influence of the comprehensive disturbance, and the compensated total control output is obtained.

9. The dual-layer composite control method for the dual three-phase motor drive-by-wire steering system of heavy-duty transport equipment according to claim 8, characterized in that, Step 3) specifically includes: An adaptive interval observer based on a superbasic function neural network is designed to compensate for disturbances in the structural model of a dual three-phase steering motor steer-by-wire system, as detailed below: The unknown disturbance in the dual three-phase motor steer-by-wire system is approximated using a superbasic function neural network, and the upper and lower bound estimates of the system disturbance are constructed as follows: ； In the formula, , and These represent the adaptive weight matrix, the upper bound of the adaptive weight matrix, and the lower bound of the adaptive weight matrix of the superbasic function neural network, respectively. , and Let represent the superbasic function, the upper bound of the superbasic function, and the lower bound of the superbasic function, respectively; , and This represents the approximation error, the upper bound of the approximation error, and the lower bound of the approximation error; This serves as the upper bound for estimating the overall system disturbance. This serves as the lower bound for estimating the overall system disturbance. The state estimation dynamic equation of the adaptive interval observer based on the superbasic function neural network can be expressed as: ； in, , These represent the upper and lower bounds of the state, respectively. , The time derivatives of the upper and lower bounds of the state; For control input; The system's measurable output vector is represented by the gain matrix of the adaptive interval observer based on the hyperbasis function neural network, which is defined as follows: ,in , , and These are the state matrix, input matrix, output matrix, and observer gain matrix of the dynamic equation for the state estimation of an adaptive interval observer based on a superbasic function neural network. and These are the upper and lower bounds of the bounded approximation error generated when a superbasic function neural network approximates an unknown perturbation; The upper and lower bound matrices of the weights in a hyperbasis function neural network are denoted as follows: and The update law is obtained from the following formula: ； In the formula, , These are the output matrices of the hidden layer of the hyperbasis function neural network. The upper and lower bounds; and Let be the upper bound weight gain matrix and the lower bound weight gain matrix of the superbasis function neural network weights, respectively, and satisfy . and The residual terms of the approximation error of the hyperbasis function neural network satisfy... and ; and This represents the upper and lower bound estimates of the output constructed by the adaptive interval observer based on a superbasic function neural network; and These are the weight decay coefficients in the upper and lower bound weight update laws, respectively. , .

10. The dual-layer composite control method for a heavy-duty transport equipment dual three-phase motor steer-by-wire system according to claim 9, characterized in that, Step 4) specifically includes: The torque tracking error of the i-th group of redundant sub-motors Defined as: ； In the formula, This is the reference torque for the i-th group of redundant sub-motors; This represents the actual output torque of the i-th redundant sub-motor; The synchronization error of each set of redundant sub-motors is defined by the difference between its actual torque and that of the other two sets of redundant sub-motors, as follows: ； In the formula, and The actual torque of the two sets of redundant sub-motors other than redundant sub-motor i; The actual torque error of redundant sub-motor i and redundant sub-motor s. The actual torque error of redundant sub-motor i and redundant sub-motor r; Torque compensation amount of the i-th group of redundant sub-motors Represented as: ； In the formula, , , These are the torque tracking errors of the three redundant sub-motors in the i, s, and r groups, respectively. and To compensate for the gain, the ratio of rotational inertia between each group of redundant sub-motors is determined. and The proportional and integral parameters of the PI controller; the average torque tracking error is defined as... ; The difference between the torque tracking error of the i-th redundant sub-motor and the torque tracking errors of the other redundant sub-motors is as follows: ； In the formula, The difference in torque tracking error between redundant sub-motor i and redundant sub-motor s. This is the difference in torque tracking error between redundant sub-motor i and redundant sub-motor r; The error between the torque tracking error of the i-th group of redundant sub-motors and the average torque tracking error of all redundant sub-motors for: ； PI parameters are dynamically adjusted based on real-time feedback; torque tracking error. Embedded state vector In this context, it is used for Actor-Critic network learning; the Actor-Critic network consists of: a policy network, a value network, and a stochastic action modifier; the policy network maps the system state to candidate PI parameters of the i-th group of redundant sub-motors: ； In the formula, Represents the candidate PI parameter vector. and These are the proportional parameter and integral parameter in the candidate PI parameter vector, respectively; The random action corrector performs random perturbation correction based on the value function of the value network, and the final execution parameters are as follows: ； In the formula, Indicates the first The actual execution PI parameter vector of the lower-level torque synchronization controller of the redundant sub-motor group; and These represent the proportional gain parameter and integral gain parameter in the actual PI parameter vector, respectively.

Images