A road feel robust control method for a six-phase permanent magnet synchronous motor steer-by-wire system

Abstract

The application discloses a kind of six-phase permanent magnet synchronous motor line control steering system road sense robust control method, based on line control steering road sense system, steps are as follows: target road sense torque is obtained by calculation;Difference is obtained by the difference between target road sense torque and actual torque;Difference is input to road sense controller, and the q-axis target current of six-phase permanent magnet synchronous motor is obtained by calculation;Unknown nonlinear disturbance of current loop is calculated by extended state observer;Six-phase reference voltage of six-phase permanent magnet synchronous motor is calculated;Six-phase reference voltage is input to six-phase permanent magnet synchronous motor, and six-phase permanent magnet synchronous motor generates torque, and is passed to steering wheel through steering column and further drives steering wheel.The application can effectively improve the robustness of road sense system, has better torque stability and robustness in healthy or fault state, maximally reduces the driving burden of driver and the steering stability of vehicle, realizes the improvement of safety and steering performance of line control steering vehicle.

Description

Technical Field

[0001] This invention belongs to the field of vehicle steer-by-wire control, specifically relating to a road feel robust control method for a six-phase permanent magnet synchronous motor steer-by-wire system. Background Technology

[0002] The safety and robustness of steer-by-wire (SbW) road feel systems across all conditions (health / fault) have always been a major concern. Six-phase permanent magnet synchronous motors (PMSMs) address this safety issue due to their redundant windings. However, when a six-phase PMSM fails, in addition to external load disturbances, internal parameter mismatches, inverter nonlinearity, cross-coupling voltage, and high-order current harmonics, severe periodic torque pulsations inevitably occur, significantly impacting performance.

[0003] Traditional robust solutions, such as current harmonic suppression, torque robust control, current robust control, external disturbance and parameter mismatch suppression, and fault-tolerant control, are all independent solutions for specific states and cannot achieve optimal performance under all conditions, including healthy and faulty states. Periodic torque pulsations occur when the windings of a six-phase permanent magnet synchronous motor are short-circuited or open-circuited. The torque control module should be able to suppress these pulsations. Compared to sliding mode control, H∞ control, and model predictive control, learning and repetitive control are more suitable solutions. However, repetitive control is ineffective against aperiodic disturbance components, and its performance relies too heavily on the system's inertial response. Stable tracking of the target road feel torque depends on good current control of the dual three-phase permanent magnet synchronous motor. Due to the rapid development of microprocessors, model predictive current control has emerged due to its simple principle and clear physical concepts, but it relies excessively on models. Model-free current predictive control algorithms avoid these drawbacks. Summary of the Invention

[0004] To address the shortcomings of the prior art, the present invention aims to provide a road feel robust control method for a six-phase permanent magnet synchronous motor steer-by-wire system, thereby solving the problem of unstable torque output in the prior art due to external load disturbances, internal parameter mismatch, inverter nonlinearity, cross-coupling voltage, and high-order current harmonics in different states of the steer-by-wire road feel system.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] The present invention discloses a road feel robust control method for a six-phase permanent magnet synchronous motor steer-by-wire system, based on a steer-by-wire road feel system, which includes: a six-phase permanent magnet synchronous motor, a steering column, a steering wheel, a steering angle sensor, a torque sensor, a road feel controller, a vehicle speed sensor, and a current sensor;

[0007] The six-phase permanent magnet synchronous motor is mechanically connected to the vehicle chassis, and its output shaft is mechanically connected to the steering column. The angle sensor and torque sensor are respectively mounted on the output shaft of the six-phase permanent magnet synchronous motor. The angle sensor collects the rotation angle of the output shaft, and the torque sensor collects the torque of the output shaft. Both the angle sensor and torque sensor are electrically connected to the road feel controller, and transmit the angle and torque signals to the road feel controller via signal lines. The road feel controller is electrically connected to the six-phase permanent magnet synchronous motor, and outputs a six-phase voltage signal, which is transmitted to the six-phase permanent magnet synchronous motor via signal lines. The six-phase permanent magnet synchronous motor generates torque and drives the steering column and steering wheel, providing road feel torque to the driver. The steering wheel is mounted on one end of the steering column. The current sensor is electrically connected to the road feel controller, and collects the current of the six-phase permanent magnet synchronous motor, transmitting it to the road feel controller via signal lines. The vehicle speed sensor collects the vehicle's speed. The steps are as follows:

[0008] 1) The vehicle speed signal collected by the vehicle speed sensor and the angle signal of the six-phase permanent magnet synchronous motor collected by the angle sensor are transmitted to the road sense controller to calculate the target road sense torque;

[0009] 2) The torque sensor collects the actual torque of the output shaft of the six-phase permanent magnet synchronous motor, and the difference between the target road-feel torque and the actual torque is obtained.

[0010] 3) Input the difference from step 2) above into the road sense controller, and calculate the target q-axis current of the six-phase permanent magnet synchronous motor through the adaptive forgetting factor iterative learning algorithm.

[0011] 4) Establish the state equation of the extended state observer of the current loop of the six-phase permanent magnet synchronous motor and discretize it. With the current and the unknown nonlinear disturbance of the current loop as state variables, the unknown nonlinear disturbance of the current loop on the d-axis, q-axis, z1-axis and z2-axis are calculated by the extended state observer.

[0012] 5) Design a superlocal model-free current predictive control algorithm, and perform discretization and beat delay processing to determine the target q-axis current of the six-phase permanent magnet synchronous motor. The input is fed into the superlocal model-free current predictive control algorithm to control the d-axis current. The target value is set to 0, and the target values ​​for the z1-axis and z2-axis currents are... and Set to 0, and modulate the target values ​​of the currents of the d-axis, q-axis, z1-axis and z2-axis into the superlocal model-free current prediction control algorithm with the unknown nonlinear disturbance input of the current loop to obtain the six-phase reference voltage of the six-phase permanent magnet synchronous motor;

[0013] 6) The six-phase reference voltage is input to the six-phase permanent magnet synchronous motor, which generates torque and transmits it to the steering wheel through the steering column, further driving the steering wheel and providing the driver with road feel torque.

[0014] Furthermore, the six-phase permanent magnet synchronous motor is mechanically connected to the vehicle chassis via bolts and nuts.

[0015] Further, step 1) specifically includes:

[0016] The vehicle speed signal collected by the vehicle speed sensor and the steering angle sensor, and the steering angle signal of the six-phase permanent magnet synchronous motor are used as inputs to the road sense controller, and the target road sense torque is calculated.

[0017] Target road feel torque includes self-aligning torque M z Steering wheel torque T caused by self-aligning torque z Represented as:

[0018]

[0019] In the formula, E(V) is the velocity coefficient; M zmax Lateral acceleration greater than or equal to 4 m / s² 2 The self-aligning torque of the front wheels at that time; a y G is the lateral acceleration of the vehicle; G is the self-alignment coefficient.

[0020] Self-aligning torque M z The following is obtained from a two-degree-of-freedom vehicle model:

[0021]

[0022] In the formula, m is the vehicle mass; u is the vehicle's longitudinal velocity; a is the distance from the front axle to the vehicle's center of gravity; b is the distance from the rear axle to the vehicle's center of gravity; C f C represents the front wheel lateral stiffness. r Rear wheel lateral stiffness; Q is front wheel load; D is tire roll offset; β′ is principal axle camber angle; l is wheelbase, l=a+b; δ f The steering angle of the vehicle's front wheels;

[0023] The damping control of road feel torque is expressed as being proportional to the steering wheel angular velocity, as follows:

[0024]

[0025] In the formula, C damp T is the damping coefficient. damp This refers to the road feel torque with damping control. The first time derivative of the steering wheel angle;

[0026] Target road feel torque Th Represented as:

[0027]

[0028] Further, step 2) specifically includes:

[0029] A torque sensor acquires the actual torque of the output shaft of a six-phase permanent magnet synchronous motor. The difference between the target road-felt torque and the actual torque is calculated to obtain the difference e between the target road-felt torque and the actual torque of the output shaft of the six-phase permanent magnet synchronous motor. T :

[0030] e T =T h -T s

[0031] In the formula, T h The target road feel torque; T s This represents the actual road feel torque.

[0032] Furthermore, step 3) specifically includes:

[0033] The learning law of the adaptive forgetting factor iterative learning algorithm is expressed as:

[0034]

[0035] In the formula, k is the number of iterations; λ is the forgetting factor, λ∈[0,1]; ρ is the control gain; L is the learning gain matrix; This is the q-axis reference current.

[0036] Furthermore, step 4) specifically includes:

[0037] Using current x1 and unknown nonlinear disturbance x2 in the current loop as state variables, the state equation of the extended state observer is:

[0038]

[0039] In the formula, z1 is the observed current value; z2 is the observed disturbance value; x1 is the current; x2 is the nonlinear disturbance; η i1 and η i2 Here are the observer parameters, where i is 1, 2, 3, and 4, where 1 represents the d-axis, 2 represents the q-axis, 3 represents the z1-axis, and 4 represents the z2-axis; α is the current coefficient; β is the voltage coefficient; u is the control input; and e is the error between the observed and actual values.

[0040] The characteristic equation of the extended state observer is:

[0041]

[0042] In the formula, A, K, and C are the parameters of the characteristic equation; s is the differential operator; I is the identity matrix; η i1 and η i2 For the discrete gain of the observer, ω0 is the bandwidth;

[0043] The state equation of the extended state observer is discretized using the forward Euler discretization method, resulting in the following discretized state equation:

[0044]

[0045] In the formula, e(k) is the difference between the observed current value z1(k) and the actual current value x1(k) at time k; z1(k+1) is the observed current value at time k+1; T is the sampling period; z2(k) is the observed current loop disturbance at time k; u(k) is the control input at time k; z2(k+1) is the observed current loop disturbance at time k+1; η i1 and η i2 For the discrete gain of the observer;

[0046] The unknown nonlinear perturbation F of the current loop along the d-axis, q-axis, z1-axis, and z2-axis is obtained by minimizing the error between the actual current value and the observed current value. d F q F z1 and F z2 .

[0047] Furthermore, step 5) specifically includes:

[0048] Using the voltage vector as the control variable and the change in stator current as the output variable, a hyperlocal model-free current prediction algorithm for the dq and z1-z2 subspaces is established as follows:

[0049]

[0050] In the formula, di d / dt is the first time derivative of the d-axis current; di q / dt is the first time derivative of the q-axis current; di z1 / dt is the first-order time derivative of the z1-axis current; di z2 / dt is the first-order time derivative of the z2-axis current; α d =α q =-R0 / L0 and α z1 =α z2 =-R0 / L z β represents the current coefficients of the dq and z1-z2 subspaces, respectively; d =β q = -1 / L0 and β z1 =β z2 =-1 / Lz F represents the voltage coefficients of the dq and z1-z2 subspaces, respectively; d F q F z1 and F z2 These are unknown nonlinear perturbations in the dq and z1-z2 subspaces, respectively.

[0051] By keeping α and β constant, the slow disturbance caused by mismatch is included in the unknown nonlinear disturbance F, and F is updated in real time to achieve dynamic current tracking and harmonic suppression.

[0052] The sampling period is set to T. The superlocal model-free current prediction algorithm is discretized using the first-order Euler equations. The discrete state equations of the six-phase permanent magnet synchronous motor are as follows:

[0053]

[0054] In the formula, and These are the predicted values ​​for the (k+1)th sampling period; k is the number of discrete iterations; u d (k) represents the d-axis input voltage; u q (k) represents the q-axis input voltage; u z1 (k) represents the z1-axis input voltage; u z2 (k) represents the z2-axis input voltage; i d i is the d-axis current; q i is the q-axis current; z1 For the z1 axis current; i z2 Z2 axis current; T is the discrete control period; α d α is the d-axis current coefficient; q α is the q-axis current coefficient; z1 α is the z1-axis current coefficient; z2 β is the z2-axis current coefficient; d β is the d-axis voltage coefficient; q β is the q-axis voltage coefficient; z1 β is the z1-axis current coefficient; z2 The z2-axis current coefficient; For d-axis target perturbation; For q-axis target perturbation; For the target disturbance along the z1 axis; The z2-axis target perturbation expansion state observer updates the data in real time for each sampling period. and

[0055] In discrete control systems, the actual voltage has a one-step delay, and the voltage equation can be expressed as:

[0056]

[0057] In the formula, k is the number of discrete iterations; u d (k) represents the d-axis input voltage; u q (k) represents the q-axis input voltage; u z1 (k) represents the z1-axis input voltage; u z2 (k) represents the z2-axis input voltage; i d i is the d-axis current; q i is the q-axis current; z1 For the z1 axis current; i z2 Z2 axis current; T is the discrete control period; α d α is the d-axis current coefficient; q α is the q-axis current coefficient; z1 α is the z1-axis current coefficient; z2 β is the z2-axis current coefficient; d β is the d-axis voltage coefficient; q β is the q-axis voltage coefficient; z1 β is the z1-axis current coefficient; z2 The z2-axis current coefficient; For d-axis target perturbation; For q-axis target perturbation; For the target disturbance along the z1 axis; The target perturbation is for the z2 axis.

[0058] Furthermore, the method also includes: when the steer-by-wire road feel system is affected by unknown external disturbances, short-circuit or open-circuit faults in the windings of the six-phase permanent magnet synchronous motor, parameter mismatch of the six-phase permanent magnet synchronous motor, dead-zone nonlinearity of the six-phase permanent magnet synchronous motor inverter, and cross-coupling voltage, unstable torque output occurs; a road feel robust control dynamic balance strategy based on a multivariable multi-objective sliding mode extremum search algorithm is established; the actual torque value collected by the torque sensor is input to the road feel controller, and the torque signal collected by the torque sensor and the actual current values ​​of the d-axis, q-axis, z1-axis, and z2-axis are input to the road feel robust control dynamic balance strategy to calculate the optimal parameters and weights; the specific steps are as follows:

[0059] 51) Input the difference between the target road feel torque and the actual road feel torque into the first cost function, minimize the difference between the target road feel torque and the actual road feel torque by designing the first sliding surface function and the second sliding surface function; and adjust the first sliding surface function and the second sliding surface function in real time by the first adaptive control law to obtain the optimal control gain and forgetting factor of the adaptive forgetting factor iterative learning algorithm.

[0060] 52) Input the differences between the target currents of the d-axis, q-axis, z1-axis, and z2-axis and the actual currents of the d-axis, q-axis, z1-axis, and z2-axis into the second cost function. Minimize the differences between the target currents of the d-axis, q-axis, z1-axis, and z2-axis and the actual currents of the d-axis, q-axis, z1-axis, and z2-axis by designing the third and fourth sliding surface functions. And adjust the third and fourth sliding surface functions in real time by the second adaptive control law to obtain the optimal dq-axis and z1-z2-axis bandwidths of the extended state observer.

[0061] 53) Establish a convex combination of the first, second, third, and fourth sliding surface functions. By minimizing the norm of the convex combination of the first, second, third, and fourth sliding surface functions, obtain the optimal weights of the first and second adaptive control laws.

[0062] Further, step 51) specifically includes:

[0063] The decision variable Ψ is designed as follows:

[0064]

[0065] In the formula, v(t)∈Rn is the control input specified in the next stage;

[0066] With the objective of minimizing the first cost function, a multivariate sliding mode extremum search algorithm is used to address the optimization problem of control gain and forgetting factor in the adaptive forgetting factor iterative learning algorithm; the first cost function is:

[0067]

[0068] In the formula, T is the sampling period of the system; e T e represents the error between the target torque and the measured torque. T =T h -T s ;

[0069] Let the decision variable Ψ1 = [λρ]; the road sense controller only measures the known cost function J1(Ψ1). Without needing to obtain the system and gradient changes in advance, the control input v(t) is designed to adjust Ψ1(t) so that J1(Ψ1) is minimized.

[0070] Design the first sliding surface function, and the second sliding surface function as follows:

[0071] σ 1i =J1(Ψ1(t))-p 1i

[0072] In the formula, p 1i<0 represents the slope of the i-th sliding surface, where i is 1 or 2;

[0073] Using σ 11 and σ 12 Optimize parameters λ and ρ respectively, and the sliding surface vector is:

[0074] σ 1i =[σ 11 σ 12 ]

[0075] Consider the driving vector as p 1i =[p 11 p 12 First adaptive control law v 1i (t) is defined as follows:

[0076]

[0077] In the formula, ξ1 is the weight related to the adaptive forgetting factor algorithm, which is obtained through step 33); sgn(·) is a 2×1 signum vector; It is a 2×1 vector. To compensate for the weak coupling assumption, The choices were limited; K g1i =diag[k g11 k g12 The control gain is a 2×2 diagonal positive definite matrix that determines the convergence rate. The sliding mode extremum search algorithm forces J1 to remain on a decreasing sliding surface vector, i.e., σ. 1i Optimization is achieved by approaching 0; the system moves towards the ideal decision variables. move.

[0078] Further, step 52) specifically includes:

[0079] With the objective of minimizing the second cost function, a multivariate sliding mode extremum search algorithm is used to handle the dq subspace bandwidth of the extended state observer. bandwidth of z1-z2 subspace The optimization problem, let the decision variables

[0080] The second cost function is:

[0081]

[0082] In the formula,

[0083] Using the third sliding surface function σ 21 and the fourth sliding surface function σ 22 Optimize bandwidth and The sliding surface vector is defined as:

[0084] σ 2j =[σ 21 σ 22 ]

[0085] In the formula, σ 2j =J2-p 2j (t) makes p 2j <0 represents the slope of the j-th sliding surface; consider the driving vector as p2 = [p 21 p2]2; j is 1 or 2;

[0086] Adjusting decision vector Minimize J2; the second adaptive law ν of the hyperlocal model-free current prediction control algorithm 2j (t) is:

[0087]

[0088] In the formula, ξ2 is the weight associated with the superlocal model-free current prediction control algorithm, which is obtained through step 33); K is a 2×1 vector; g2j =diag[k g21 k g22 The control gain, a 2×2 diagonal positive definite matrix, determines the convergence speed; the sliding mode extremum search algorithm forces J2 to remain on the decreasing sliding surface vector, i.e., σ. 2j Optimization is achieved by approaching 0; the system moves toward the optimal decision variable. move.

[0089] Furthermore, step 53) specifically includes:

[0090] The weights of the first adaptive control law are ξ1 = ν, and the weights of the second adaptive control law are ξ2 = 1 - ν; the multi-objective problem is transformed into a single-objective function; where ν is the value obtained through multi-objective sliding mode extremum search; the convex combination ω of the sliding mode components is:

[0091]

[0092] In the formula, ω represents the convex combination of sliding mode vectors sgn(sin(·)); ω1 represents the convex combination of sliding mode vectors sgn(sin(·)) of the first adaptive control law; ω2 represents the convex combination of sliding mode vectors sgn(sin(·)) of the second adaptive control law; σ 11 σ is the first sliding surface function; 12 σ is the second sliding surface function; 21 σ is the third sliding surface function; 22 γ is the fourth sliding surface function; 11 γ12 γ 21 and γ 22 This is the 2×1 vector corresponding to the sliding surface function;

[0093] Using the Frobenius norm, we obtain ||ω||:

[0094]

[0095] By obtaining the weights for the multi-objective problem, we can obtain the minimum norm ||ω||, and thus obtain the optimal weights.

[0096] Furthermore, step 6) specifically includes:

[0097] The six-phase permanent magnet synchronous motor uses vector space decoupling (VSD) for coordinate transformation, decomposing the stationary coordinate system into three mutually orthogonal subspaces: dq, z1-z2, and o1-o2. In the synchronous rotating coordinate system, the voltage equation is expressed as:

[0098]

[0099] In the formula, u d u q Represents the stator voltage component of the dq subspace; u z1 u z2 Represents the stator voltage components of the z1-z2 subspace; i d i q Represents the stator current in the dq subspace; i z1 i z2 R0 represents the stator current in the z1-z2 subspace; R0 is the stator resistance; the inductance and leakage inductance in the dq subspace are represented by L0 and L... z ;ω e Represents electric velocity; ψ f This represents the magnetic flux of a permanent magnet;

[0100] The motion and torque equations of a six-phase permanent magnet synchronous motor are as follows:

[0101]

[0102]

[0103] In the formula, ω m J is the mechanical angular velocity. m T represents the moment of inertia of the induction motor. e For electromagnetic torque; B m T is the damping coefficient of the road feeler motor. l P is the load torque; n For extreme logarithms; i q f is the q-axis current; mis the Coulomb friction coefficient.

[0104] This invention eliminates the need to design separate solutions for each state, is not model-based, and does not require prior acquisition of disturbance uncertainty boundaries and fault information. In road-sensing robust control, it can track the system state and dynamically adjust control parameters and weights to ensure that the road-sensing system can achieve optimal robust performance under various states.

[0105] The beneficial effects of this invention are:

[0106] This invention can effectively improve the robustness of the road feel system, providing better torque stability and robustness under healthy or fault conditions, minimizing the driver's operational burden and vehicle handling stability, and thus improving the safety and handling performance of steer-by-wire vehicles. Therefore, it has broad market application prospects. Attached Figure Description

[0107] Figure 1 This is a schematic diagram of the control principle of the road-sensing robust control method in this invention.

[0108] Figure 2 This is a structural diagram of the adaptive forgetting factor iterative learning algorithm in this invention.

[0109] Figure 3 This is a structural diagram of the superlocal model-free current prediction control algorithm with an extended state observer in this invention.

[0110] Figure 4 This is a schematic diagram of the adaptive forgetting factor iterative learning algorithm with multivariate sliding mode extremum search algorithm in this invention.

[0111] Figure 5 This is a schematic diagram of the hyperlocal model-free current prediction control algorithm with multivariable sliding mode extremum search algorithm in this invention.

[0112] Figure 6 This is a schematic diagram of the multi-objective sliding mode extremum search algorithm in this invention. Detailed Implementation

[0113] 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.

[0114] Reference Figures 1 to 6 As shown, the present invention discloses a road feel robust control method for a six-phase permanent magnet synchronous motor steer-by-wire system, based on a steer-by-wire road feel system, which includes: a six-phase permanent magnet synchronous motor, a steering column, a steering wheel, a steering angle sensor, a torque sensor, a road feel controller, a vehicle speed sensor, and a current sensor.

[0115] The six-phase permanent magnet synchronous motor is mechanically connected to the vehicle chassis (in this example, via bolts and nuts), and its output shaft is mechanically connected to the steering column. The angle sensor and torque sensor are respectively mounted on the output shaft of the six-phase permanent magnet synchronous motor. The angle sensor collects the rotation angle of the output shaft, and the torque sensor collects the torque of the output shaft. Both the angle sensor and torque sensor are electrically connected to the road feel controller, and transmit the angle and torque signals to the road feel controller via signal lines. The road feel controller is electrically connected to the six-phase permanent magnet synchronous motor, and outputs a six-phase voltage signal, which is transmitted to the six-phase permanent magnet synchronous motor via signal lines. The six-phase permanent magnet synchronous motor generates torque and drives the steering column and steering wheel, providing road feel torque to the driver. The steering wheel is mounted on one end of the steering column. The current sensor is electrically connected to the road feel controller, and collects the current of the six-phase permanent magnet synchronous motor, transmitting it to the road feel controller via signal lines. The vehicle speed sensor collects the vehicle's speed. The steps are as follows:

[0116] 1) The vehicle speed signal collected by the vehicle speed sensor and the angle signal of the six-phase permanent magnet synchronous motor collected by the angle sensor are transmitted to the road sense controller to calculate the target road sense torque; specifically including:

[0117] The vehicle speed signal collected by the vehicle speed sensor and the steering angle sensor, and the steering angle signal of the six-phase permanent magnet synchronous motor are used as inputs to the road sense controller, and the target road sense torque is calculated.

[0118] Target road feel torque includes self-aligning torque M z Steering wheel torque T caused by self-aligning torque z Represented as:

[0119]

[0120] In the formula, E(V) is the velocity coefficient; M zmax Lateral acceleration greater than or equal to 4 m / s² 2 The self-aligning torque of the front wheels at that time; a y G is the lateral acceleration of the vehicle; G is the self-alignment coefficient.

[0121] Self-aligning torque M z The following is obtained from a two-degree-of-freedom vehicle model:

[0122]

[0123] In the formula, m is the vehicle mass; u is the vehicle's longitudinal velocity; a is the distance from the front axle to the vehicle's center of gravity; b is the distance from the rear axle to the vehicle's center of gravity; C f C represents the front wheel lateral stiffness. rRear wheel lateral stiffness; Q is front wheel load; D is tire roll offset; β′ is principal axle camber angle; l is wheelbase, l=a+b; δ f The steering angle of the vehicle's front wheels;

[0124] The damping control of road feel torque is expressed as being proportional to the steering wheel angular velocity, as follows:

[0125]

[0126] In the formula, C damp T is the damping coefficient. damp This refers to the road feel torque with damping control. The first time derivative of the steering wheel angle;

[0127] Target road feel torque T h Represented as:

[0128] T h =T z +T damp .

[0129] 2) The torque sensor acquires the actual torque of the output shaft of the six-phase permanent magnet synchronous motor, and the difference between the target road-felt torque and the actual torque is obtained; specifically including:

[0130] A torque sensor acquires the actual torque of the output shaft of a six-phase permanent magnet synchronous motor. The difference between the target road-felt torque and the actual torque is calculated to obtain the difference e between the target road-felt torque and the actual torque of the output shaft of the six-phase permanent magnet synchronous motor. T :

[0131] e T =T h -T s

[0132] In the formula, T h The target road feel torque; T s This represents the actual road feel torque.

[0133] 3) Input the difference from step 2) above into the road sense controller, and calculate the target q-axis current of the six-phase permanent magnet synchronous motor through the adaptive forgetting factor iterative learning algorithm. Specifically, it includes:

[0134] The learning law of the adaptive forgetting factor iterative learning algorithm is expressed as:

[0135]

[0136] In the formula, k is the number of iterations; λ is the forgetting factor, λ∈[0,1]; ρ is the control gain; L is the learning gain matrix; This is the q-axis reference current.

[0137] 4) Establish the state equations of the extended state observer for the current loop of the six-phase permanent magnet synchronous motor, and discretize them. Using the current and the unknown nonlinear disturbance of the current loop as state variables, calculate the unknown nonlinear disturbances of the current loop along the d-axis, q-axis, z1-axis, and z2-axis through the extended state observer; specifically including:

[0138] Using current x1 and unknown nonlinear disturbance x2 in the current loop as state variables, the state equation of the extended state observer is:

[0139]

[0140] In the formula, z1 is the observed current value; z2 is the observed disturbance value; x1 is the current; x2 is the nonlinear disturbance; η i1 and η i2 Here are the observer parameters, where i is 1, 2, 3, and 4, where 1 represents the d-axis, 2 represents the q-axis, 3 represents the z1-axis, and 4 represents the z2-axis; α is the current coefficient; β is the voltage coefficient; u is the control input; and e is the error between the observed and actual values.

[0141] The characteristic equation of the extended state observer is:

[0142]

[0143] In the formula, A, K, and C are the parameters of the characteristic equation; s is the differential operator; I is the identity matrix; η i1 and η i2 For the discrete gain of the observer, ω0 is the bandwidth;

[0144] The state equation of the extended state observer is discretized using the forward Euler discretization method, resulting in the following discretized state equation:

[0145]

[0146] In the formula, e(k) is the difference between the observed current value z1(k) and the actual current value x1(k) at time k; z1(k+1) is the observed current value at time k+1; T is the sampling period; z2(k) is the observed current loop disturbance at time k; u(k) is the control input at time k; z2(k+1) is the observed current loop disturbance at time k+1; η i1 and η i2 For the discrete gain of the observer;

[0147] The unknown nonlinear perturbation F of the current loop along the d-axis, q-axis, z1-axis, and z2-axis is obtained by minimizing the error between the actual current value and the observed current value. d F q F z1 and F z2 .

[0148] 5) Design a superlocal model-free current predictive control algorithm, and perform discretization and beat delay processing to determine the target q-axis current of the six-phase permanent magnet synchronous motor. The input is fed into the superlocal model-free current predictive control algorithm to control the d-axis current. The target value is set to 0, and the target values ​​for the z1-axis and z2-axis currents are... and Setting it to 0, the target values ​​of the d-axis, q-axis, z1-axis, and z2-axis currents are modulated into the superlocal model-free current predictive control algorithm with unknown nonlinear disturbance input to the current loop to obtain the six-phase reference voltage of the six-phase permanent magnet synchronous motor; specifically including:

[0149] Using the voltage vector as the control variable and the change in stator current as the output variable, a hyperlocal model-free current prediction algorithm for the dq and z1-z2 subspaces is established as follows:

[0150]

[0151] In the formula, di d / dt is the first time derivative of the d-axis current; di q / dt is the first time derivative of the q-axis current; di z1 / dt is the first-order time derivative of the z1-axis current; di z2 / dt is the first-order time derivative of the z2-axis current; α d =α q =-R0 / L0 and α z1 =α z2 =-R0 / L z β represents the current coefficients of the dq and z1-z2 subspaces, respectively; d =β q = -1 / L0 and β z1 =β z2 =-1 / L z F represents the voltage coefficients of the dq and z1-z2 subspaces, respectively; d F q F z1 and F z2 These are unknown nonlinear perturbations in the dq and z1-z2 subspaces, respectively.

[0152] By keeping α and β constant, the slow disturbance caused by mismatch is included in the unknown nonlinear disturbance F, and F is updated in real time to achieve dynamic current tracking and harmonic suppression.

[0153] The sampling period is set to T. The superlocal model-free current prediction algorithm is discretized using the first-order Euler equations. The discrete state equations of the six-phase permanent magnet synchronous motor are as follows:

[0154]

[0155] In the formula, and These are the predicted values ​​for the (k+1)th sampling period; k is the number of discrete iterations; u d (k) represents the d-axis input voltage; u q (k) represents the q-axis input voltage; u z1 (k) represents the z1-axis input voltage; u z2 (k) represents the z2-axis input voltage; i d i is the d-axis current; q i is the q-axis current; z1 For the z1 axis current; i z2 Z2 axis current; T is the discrete control period; α d α is the d-axis current coefficient; q α is the q-axis current coefficient; z1 α is the z1-axis current coefficient; z2 β is the z2-axis current coefficient; d β is the d-axis voltage coefficient; q β is the q-axis voltage coefficient; z1 β is the z1-axis current coefficient; z2 The z2-axis current coefficient; For d-axis target perturbation; For q-axis target perturbation; For the target disturbance along the z1 axis; The z2-axis target perturbation expansion state observer updates the data in real time for each sampling period. and

[0156] In discrete control systems, the actual voltage has a one-step delay, and the voltage equation can be expressed as:

[0157]

[0158] In the formula, k is the number of discrete iterations; u d (k) represents the d-axis input voltage; u q (k) represents the q-axis input voltage; u z1 (k) represents the z1-axis input voltage; u z2 (k) represents the z2-axis input voltage; i d i is the d-axis current; q i is the q-axis current; z1 For the z1 axis current; i z2 Z2 axis current; T is the discrete control period; α d α is the d-axis current coefficient; q α is the q-axis current coefficient; z1 α is the z1-axis current coefficient; z2β is the z2-axis current coefficient; d β is the d-axis voltage coefficient; q β is the q-axis voltage coefficient; z1 β is the z1-axis current coefficient; z2 The z2-axis current coefficient; For d-axis target perturbation; For q-axis target perturbation; For the target disturbance along the z1 axis; The target perturbation is for the z2 axis.

[0159] When the steer-by-wire road feel system is affected by unknown external disturbances, short-circuit or open-circuit faults in the windings of the six-phase permanent magnet synchronous motor, parameter mismatch of the six-phase permanent magnet synchronous motor, dead-zone nonlinearity of the six-phase permanent magnet synchronous motor inverter, and cross-coupling voltage, unstable torque output occurs. A road feel robust control dynamic balance strategy based on a multivariable, multi-objective sliding mode extremum search algorithm is established. The actual torque value collected by the torque sensor is input to the road feel controller. The torque signal collected by the torque sensor, along with the actual current values ​​of the d-axis, q-axis, z1-axis, and z2-axis, are input to the road feel robust control dynamic balance strategy to calculate the optimal parameters and weights. The specific steps are as follows:

[0160] 51) Input the difference between the target road feel torque and the actual road feel torque into the first cost function, minimize the difference between the target road feel torque and the actual road feel torque by designing the first sliding surface function and the second sliding surface function; and adjust the first sliding surface function and the second sliding surface function in real time by the first adaptive control law to obtain the optimal control gain and forgetting factor of the adaptive forgetting factor iterative learning algorithm.

[0161] 52) Input the differences between the target currents of the d-axis, q-axis, z1-axis, and z2-axis and the actual currents of the d-axis, q-axis, z1-axis, and z2-axis into the second cost function. Minimize the differences between the target currents of the d-axis, q-axis, z1-axis, and z2-axis and the actual currents of the d-axis, q-axis, z1-axis, and z2-axis by designing the third and fourth sliding surface functions. And adjust the third and fourth sliding surface functions in real time by the second adaptive control law to obtain the optimal dq-axis and z1-z2-axis bandwidths of the extended state observer.

[0162] 53) Establish a convex combination of the first, second, third, and fourth sliding surface functions. By minimizing the norm of the convex combination of the first, second, third, and fourth sliding surface functions, obtain the optimal weights of the first and second adaptive control laws.

[0163] Specifically, step 51) includes:

[0164] The decision variable Ψ is designed as follows:

[0165]

[0166] In the formula, v(t)∈R n It is the control input specified in the next stage;

[0167] With the objective of minimizing the first cost function, a multivariate sliding mode extremum search algorithm is used to address the optimization problem of control gain and forgetting factor in the adaptive forgetting factor iterative learning algorithm; the first cost function is:

[0168]

[0169] In the formula, T is the sampling period of the system; e T e represents the error between the target torque and the measured torque. T =T h -T s ;

[0170] Let the decision variable Ψ1 = [λρ]; the road sense controller only measures the known cost function J1(Ψ1). Without needing to obtain the system and gradient changes in advance, the control input v(t) is designed to adjust Ψ1(t) so that J1(Ψ1) is minimized.

[0171] Design the first sliding surface function, and the second sliding surface function as follows:

[0172] σ 1i =J1(Ψ1(t))-p 1i

[0173] In the formula, p 1i <0 represents the slope of the i-th sliding surface, where i is 1 or 2;

[0174] Using σ 11 and σ 12 Optimize parameters λ and ρ respectively, and the sliding surface vector is:

[0175] σ 1i =[σ 11 σ 12 ]

[0176] Consider the driving vector as p 1i =[p 11 p 12 First adaptive control law v 1i (t) is defined as follows:

[0177]

[0178] In the formula, ξ1 is the weight related to the adaptive forgetting factor algorithm, which is obtained through step 33); sgn(·) is a 2×1 signum vector; It is a 2×1 vector. To compensate for the weak coupling assumption, The choices were limited; K g1i =diag[k g11 k g12 The control gain is a 2×2 diagonal positive definite matrix that determines the convergence rate. The sliding mode extremum search algorithm forces J1 to remain on a decreasing sliding surface vector, i.e., σ. 1i Optimization is achieved by approaching 0; the system moves towards the ideal decision variables. move.

[0179] Specifically, step 52) includes:

[0180] With the objective of minimizing the second cost function, a multivariate sliding mode extremum search algorithm is used to handle the dq subspace bandwidth of the extended state observer. bandwidth of z1-z2 subspace The optimization problem, let the decision variables

[0181] The second cost function is:

[0182]

[0183] In the formula,

[0184] Using the third sliding surface function σ 21 and the fourth sliding surface function σ 22 Optimize bandwidth and The sliding surface vector is defined as:

[0185] σ 2j =[σ 21 σ 22 ]

[0186] In the formula, σ 2j =J2-p 2j (t) makes p 2j <0 represents the slope of the j-th sliding surface; consider the driving vector as p2 = [p 21 p2]2; j is 1 or 2;

[0187] Adjusting decision vector Minimize J2; the second adaptive law ν of the hyperlocal model-free current prediction control algorithm 2j (t) is:

[0188]

[0189] In the formula, ξ2 is the weight associated with the superlocal model-free current prediction control algorithm, which is obtained through step 33); K is a 2×1 vector; g2j =diag[k g21 k g22 The control gain, a 2×2 diagonal positive definite matrix, determines the convergence speed; the sliding mode extremum search algorithm forces J2 to remain on the decreasing sliding surface vector, i.e., σ. 2j Optimization is achieved by approaching 0; the system moves toward the optimal decision variable. move.

[0190] Specifically, step 53) includes:

[0191] The weights of the first adaptive control law are ξ1 = ν, and the weights of the second adaptive control law are ξ2 = 1 - ν; the multi-objective problem is transformed into a single-objective function; where ν is the value obtained through multi-objective sliding mode extremum search; the convex combination ω of the sliding mode components is:

[0192]

[0193] In the formula, ω represents the convex combination of sliding mode vectors sgn(sin(·)); ω1 represents the convex combination of sliding mode vectors sgn(sin(·)) of the first adaptive control law; ω2 represents the convex combination of sliding mode vectors sgn(sin(·)) of the second adaptive control law; σ 11 σ is the first sliding surface function; 12 σ is the second sliding surface function; 21 σ is the third sliding surface function; 22 γ is the fourth sliding surface function; 11 γ 12 γ 21 and γ 22 This is the 2×1 vector corresponding to the sliding surface function;

[0194] Using the Frobenius norm, we obtain ||ω||:

[0195]

[0196] By obtaining the weights for the multi-objective problem, we can obtain the minimum norm ||ω||, and thus obtain the optimal weights.

[0197] 6) A six-phase reference voltage is input to the six-phase permanent magnet synchronous motor, which generates torque. This torque is transmitted to the steering wheel through the steering column and further drives the steering wheel, providing the driver with road feel torque; specifically including:

[0198] The six-phase permanent magnet synchronous motor uses vector space decoupling (VSD) for coordinate transformation, decomposing the stationary coordinate system into three mutually orthogonal subspaces: dq, z1-z2, and o1-o2. In the synchronous rotating coordinate system, the voltage equation is expressed as:

[0199]

[0200] In the formula, u d u q Represents the stator voltage component of the dq subspace; u z1 u z2 Represents the stator voltage components of the z1-z2 subspace; i d i q Represents the stator current in the dq subspace; i z1 i z2 R0 represents the stator current in the z1-z2 subspace; R0 is the stator resistance; the inductance and leakage inductance in the dq subspace are represented by L0 and L... z ;ω e Represents electric velocity; ψ f This represents the magnetic flux of a permanent magnet;

[0201] The motion and torque equations of a six-phase permanent magnet synchronous motor are as follows:

[0202]

[0203] In the formula, ω m J is the mechanical angular velocity. m T represents the moment of inertia of the induction motor. e For electromagnetic torque; B m T is the damping coefficient of the road feeler motor. l P is the load torque; n For extreme logarithms; i q f is the q-axis current; m is the Coulomb friction coefficient.

[0204] 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 road-feel-robust control method for a six-phase permanent magnet synchronous motor steer-by-wire system, characterized in that, Based on the steer-by-wire road feel system, the steer-by-wire road feel system includes: a six-phase permanent magnet synchronous motor, a steering column, a steering wheel, a steering angle sensor, a torque sensor, a road feel controller, a vehicle speed sensor, and a current sensor; The six-phase permanent magnet synchronous motor is mechanically connected to the vehicle chassis, and its output shaft is mechanically connected to the steering column. Angle and torque sensors are respectively mounted on the output shaft of the six-phase permanent magnet synchronous motor. The angle sensor collects the rotation angle of the output shaft, and the torque sensor collects the torque of the output shaft. Both the angle and torque sensors are electrically connected to the road feel controller and send the angle and torque signals to it. The road feel controller is electrically connected to the six-phase permanent magnet synchronous motor, outputting a six-phase voltage signal and sending it to the motor. The six-phase permanent magnet synchronous motor generates torque and drives the steering column and steering wheel. The steering wheel is mounted on one end of the steering column. The current sensor is electrically connected to the road feel controller, collecting the current of the six-phase permanent magnet synchronous motor and transmitting it to the controller. The vehicle speed sensor collects the vehicle's speed. The steps are as follows: 1) The vehicle speed signal collected by the vehicle speed sensor and the rotation angle signal of the six-phase permanent magnet synchronous motor collected by the rotation angle sensor are transmitted to the road sense controller to calculate the target road sense torque; 2) The torque sensor collects the actual torque of the output shaft of the six-phase permanent magnet synchronous motor, and the difference between the target road-felt torque and the actual torque is obtained. 3) Input the difference from step 2) above into the road sensor controller to calculate the six-phase permanent magnet synchronous motor. Target current of shaft ; 4) Establish the state equations of the extended state observer for the current loop of the six-phase permanent magnet synchronous motor, and discretize them. Using the current and the unknown nonlinear disturbance of the current loop as state variables, calculate the state equations using the extended state observer. axis, axis, shaft and The current loop of the shaft has an unknown nonlinear disturbance; 5) Design a superlocal model-free current predictive control algorithm, and perform discretization and beat delay processing on the six-phase permanent magnet synchronous motor. Target current of shaft The input is fed into the superlocal model-free current predictive control algorithm, shaft current The target value is set to 0. shaft and Target value of shaft current and Set to 0, axis, axis, shaft and In the superlocal model-free current prediction control algorithm with unknown nonlinear disturbance input to the shaft current target value and the current loop unknown, the six-phase reference voltage of the six-phase permanent magnet synchronous motor is obtained. 6) The six-phase reference voltage is input to the six-phase permanent magnet synchronous motor, which generates torque and transmits it to the steering wheel through the steering column, further driving the steering wheel.

2. The road feel robust control method for a six-phase permanent magnet synchronous motor steer-by-wire system according to claim 1, characterized in that, Step 1) specifically includes: The vehicle speed signal collected by the vehicle speed sensor and the steering angle sensor, and the steering angle signal of the six-phase permanent magnet synchronous motor are used as inputs to the road sense controller, and the target road sense torque is calculated. Target road feel torque includes self-aligning torque Steering wheel torque caused by self-aligning torque Represented as: ； In the formula, The velocity coefficient; Lateral acceleration greater than 4 m / s² 2 The self-aligning torque of the front wheels at that time; This refers to the vehicle's lateral acceleration. The self-alignment coefficient; Self-aligning torque The following is obtained from a two-degree-of-freedom vehicle model: ； In the formula, For vehicle quality; The longitudinal speed of the vehicle; This is the distance from the front axle of the vehicle to the vehicle's center of gravity. This is the distance from the rear axle of the vehicle to the vehicle's center of gravity. This refers to the front wheel lateral stiffness. Rear wheel lateral stiffness; For front wheel load; This refers to the tire roll offset. Principal axis tilt angle; Wheelbase ; The steering angle of the vehicle's front wheels; The damping control of road feel torque is expressed as being proportional to the steering wheel angular velocity, as follows: ； In the formula, The damping coefficient is... This refers to the road feel torque with damping control. The first time derivative of the steering wheel angle; Target road feel torque Represented as: 。 3. The road feel robust control method for a six-phase permanent magnet synchronous motor steer-by-wire system according to claim 2, characterized in that, Step 2) specifically includes: A torque sensor acquires the actual torque of the output shaft of a six-phase permanent magnet synchronous motor. The difference between the target road-feel torque and the actual torque is then calculated to obtain the difference between the target road-feel torque and the actual torque of the output shaft of the six-phase permanent magnet synchronous motor. : ； In the formula, The target road feel torque; This represents the actual road feel torque.

4. The road feel robust control method for a six-phase permanent magnet synchronous motor steer-by-wire system according to claim 3, characterized in that, Step 3) specifically includes: The learning law of the adaptive forgetting factor iterative learning algorithm is expressed as: ； In the formula, This represents the number of iterations. Forgetting factor, ; To control the gain; To learn the gain matrix; for Target current of the shaft.

5. The road feel robust control method for a six-phase permanent magnet synchronous motor steer-by-wire system according to claim 4, characterized in that, Step 4) specifically includes: With current With unknown nonlinear disturbance of current loop As state variables, the state equation of the extended state observer is: ； In the formula, These are current observations; These are perturbation observations; For current; It is a nonlinear disturbance; and For observer parameters, The numbers are 1, 2, 3, and 4, where 1 represents... Axis, 2 represents Axis, 3 represents Axis, 4 represents axis; The current coefficient; Voltage coefficient; For control input; This represents the error between the observed and actual values. The characteristic equation of the extended state observer is: ； In the formula, , and These are the parameters of the characteristic equation; It is a differential operator; It is the identity matrix; and For the discrete gain of the observer, , ; For bandwidth; The state equation of the extended state observer is discretized using the forward Euler discretization method, resulting in the following discretized state equation: ； In the formula, For the first Current observation at time and actual current value The difference; For the first Current observation at any given time; The sampling period; For the first Observations of current loop disturbance at any given time; For the first Constantly control input; For the first Observations of current loop disturbance at any given time; and For the discrete gain of the observer; By minimizing the error between the actual current value and the observed current value, we obtain... axis, axis, shaft and Unknown nonlinear disturbance in the current loop of the shaft , , and .

6. The road feel robust control method for a six-phase permanent magnet synchronous motor steer-by-wire system according to claim 5, characterized in that, Step 5) specifically includes: Using the voltage vector as the control variable and the change in stator current as the output variable, a system is established. and The hyperlocal model-free current prediction algorithm for subspaces is as follows: ； In the formula, for The first time derivative of the shaft current; for The first time derivative of the shaft current; for The first time derivative of the shaft current; for The first time derivative of the shaft current; and They are respectively , The current coefficient of the subspace; and They are respectively , Voltage coefficient of subspace; , , and They are respectively , Unknown nonlinear perturbation in subspace; set up and The slow perturbations caused by mismatch are kept constant and included in the unknown nonlinear perturbations. In China, updates in real time Achieve dynamic current tracking and harmonic suppression; Setting the sampling period is The hyperlocal model-free current prediction algorithm discretizes the state using the first-order Euler equations. The discrete state equations of the six-phase permanent magnet synchronous motor are as follows: ； In the formula, , , and The first Predicted values ​​for each sampling period; The number of discrete iterations; for Shaft input voltage; for Shaft input voltage; for Shaft input voltage; for Shaft input voltage; for shaft current; for shaft current; for shaft current; for shaft current; For discrete control cycles; for Shaft current coefficient; for Shaft current coefficient; for Shaft current coefficient; for Shaft current coefficient; for Shaft voltage coefficient; for Shaft voltage coefficient; for Shaft current coefficient; for Shaft current coefficient; for Axis target disturbance; for Axis target disturbance; for Axis target disturbance; for The axis target perturbation expansion state observer updates the data in real time for each sampling period. , , and ; In discrete control systems, the actual voltage has a one-step delay, and the voltage equation can be expressed as: ； In the formula, The number of discrete iterations; for Shaft input voltage; for Shaft input voltage; for Shaft input voltage; for Shaft input voltage; for shaft current; for shaft current; for shaft current; for shaft current; For discrete control cycles; for Shaft current coefficient; for Shaft current coefficient; for Shaft current coefficient; for Shaft current coefficient; for Shaft voltage coefficient; for Shaft voltage coefficient; for Shaft current coefficient; for Shaft current coefficient; for Axis target disturbance; for Axis target disturbance; for Axis target disturbance; for Axis target disturbance.

7. The road feel robust control method for a six-phase permanent magnet synchronous motor steer-by-wire system according to claim 6, characterized in that, The method further includes: when the steer-by-wire road feel system is subjected to unknown external disturbances, short-circuit or open-circuit faults in the windings of the six-phase permanent magnet synchronous motor, parameter mismatch of the six-phase permanent magnet synchronous motor, dead-zone nonlinearity of the six-phase permanent magnet synchronous motor inverter, and cross-coupling voltage, unstable torque output occurs; a road feel robust control dynamic balance strategy based on a multivariable multi-objective sliding mode extreme value search algorithm is established; the actual torque value collected by the torque sensor is input to the road feel controller, and the torque signal collected by the torque sensor is compared with... axis, axis, shaft and The actual axle current value is input into the road feel robust control dynamic balancing strategy to calculate the optimal parameters and weights; the specific steps are as follows: 51) Input the difference between the target road feel torque and the actual road feel torque into the first cost function, minimize the difference between the target road feel torque and the actual road feel torque by designing the first sliding surface function and the second sliding surface function; and adjust the first sliding surface function and the second sliding surface function in real time by the first adaptive control law to obtain the optimal control gain and forgetting factor of the adaptive forgetting factor iterative learning algorithm. 52) will axis, axis, shaft and Target current of shaft and axis, axis, shaft and The difference between the actual shaft currents is input into the second cost function, and minimized by designing the third and fourth sliding surface functions. axis, axis, shaft and Target current of shaft and axis, axis, shaft and The difference between the actual shaft currents; and by adjusting the third and fourth sliding surface functions in real time using the second adaptive control law, the optimal value of the extended state observer is obtained. shaft and Axis bandwidth; 53) Establish a convex combination of the first, second, third, and fourth sliding surface functions. By minimizing the norm of the convex combination of the first, second, third, and fourth sliding surface functions, obtain the optimal weights of the first and second adaptive control laws.

8. The road feel robust control method for a six-phase permanent magnet synchronous motor steer-by-wire system according to claim 7, characterized in that, Step 51) specifically includes: Decision variables Designed as follows: ； In the formula, It is the control input specified in the next stage; With the objective of minimizing the first cost function, a multivariate sliding mode extremum search algorithm is used to address the optimization problem of control gain and forgetting factor in the adaptive forgetting factor iterative learning algorithm; the first cost function is: ； In the formula, The sampling period of the system; The error between the target torque and the measured torque. ; Let decision variables The road sensor controller only measures known cost functions. Design control inputs without prior knowledge of system and gradient changes. Thus regulating , make minimize; Design the first sliding surface function, and the second sliding surface function as follows: ； In the formula, For the first The slope of each sliding surface, It can be 1 or 2; use and Optimize parameters separately and The sliding mode surface vector is: ； Consider the driving vector as First adaptive control law The definition is as follows: ； In the formula, Weights associated with the adaptive forgetting factor algorithm; It is a 2×1 signum vector; It is a 2×1 vector. In order to compensate for the weak coupling assumption, The choices were limited; To control the gain, a 2×2 diagonal positive definite matrix is ​​used, which determines the convergence rate; the sliding mode extremum search algorithm forces... Maintained on the decreasing sliding surface vector, i.e. Optimization is achieved by approaching 0; the system moves towards the ideal decision variables. move.

9. The road feel robust control method for a six-phase permanent magnet synchronous motor steer-by-wire system according to claim 8, characterized in that, Step 52) specifically includes: With the objective of minimizing the second cost function, a multivariate sliding mode extremum search algorithm is used to process the extended state observer. Subspace bandwidth and Subspace bandwidth The optimization problem, let the decision variables ; The second cost function is: ； In the formula, ; ; ; ; Using the third sliding surface function and the fourth sliding surface function Optimize bandwidth and The sliding mode surface vector is defined as: ； In the formula, , making For the first The slope of each sliding surface; considering the driving vector as ; It can be 1 or 2; Adjusting decision vector , making Minimize; the second adaptive law of the hyperlocal model-free current prediction control algorithm for: ； In the formula, The weights are those related to the hyperlocal model-free current prediction control algorithm; It is a 2×1 vector; The control gain, a 2×2 diagonal positive definite matrix, determines the convergence rate; the sliding mode extremum search algorithm uses forced... Maintain on the decreasing sliding surface vector, i.e. Optimization is achieved by approaching 0; the system moves toward the optimal decision variable. move.

10. The road feel robust control method for a six-phase permanent magnet synchronous motor steer-by-wire system according to claim 9, characterized in that, Step 53) specifically includes: The weights of the first adaptive control law are The weights of the second adaptive control law are Transform the multi-objective problem into a single-objective function; where The value is obtained through multi-objective sliding mode extremum search; convex combination of sliding mode components. for: ； In the formula, Represents the sliding mode vector convex combination; The sliding mode vector representing the first adaptive control law convex combination; The sliding mode vector representing the second adaptive control law convex combination; This is the function of the first sliding surface; For the second sliding surface function; For the third sliding surface function; This is the function of the fourth sliding surface; , , and This is the 2×1 vector corresponding to the sliding surface function; Using the Frobenius norm, we obtain : ； To obtain the weights of a multi-objective problem, we obtain the minimum norm. This allows us to obtain the optimal weights.

Images