A four-wheel independent drive electric vehicle wheel hub motor fault diagnosis method

Abstract

This invention discloses a fault diagnosis method for hub motors in four-wheel independently driven electric vehicles. A seven-degree-of-freedom vehicle dynamics model and a hub motor fault model are established. A hub motor fault isolation method based on the plane projection theorem is proposed, which sorts the angles between the steady-state solution of the input residual information and the fault feature vector to obtain the fault isolation signal, overcoming the shortcomings of existing methods such as model rigidity and poor real-time performance. A robust H∞ fault estimator considering vehicle speed perturbations and external interference uncertainties is constructed, minimizing the residual values ​​of fault feature signals and estimation coefficients. This invention provides effective information for the chassis control layer, reducing the risks of sensor and onboard component aging and circuit failures. The proposed fault detection, fault isolation, and fault estimation methods not only handle vehicle speed perturbations and parameter uncertainties during actual vehicle operation but also significantly reduce external interference, exhibiting good accuracy and fault tolerance, thus improving the safety and stability of electric vehicles.

Description

Technical Field

[0001] This invention belongs to the field of new energy vehicle design and manufacturing. It designs a fault isolation and fault diagnosis method based on the overdrive architecture of a four-wheel independent drive electric vehicle, which significantly improves the reliability and safety of the vehicle system. Background Technology

[0002] Four-wheel independent drive electric vehicles (EVs) are poised to become a highly promising automotive architecture due to their advanced intelligent domain controllers, efficient use of renewable energy, and precise allocation of overdrive controllable units. EVs can utilize an in-wheel (or hub) motor to control traction or braking force at a specific wheel, offering significant advantages such as simplified chassis structure, rapid dynamic response, and accurate control execution. However, overdrive systems also pose significant safety risks. The instability of purely electric control systems can lead to in-wheel (or hub) motor failure or even malfunction, resulting in serious traffic accidents. Therefore, to ensure the vehicle can return to normal operation under any circumstances when the vehicle actuators experience such conditions, it is crucial to design real-time, robust fault isolation and diagnosis methods to guarantee the system's safety and stability.

[0003] However, for four-wheel independent drive electric vehicles (EVs), traditional active safety control systems are insufficient to effectively prevent and control multiple subsystems from malfunctioning. Because each system in a distributed four-wheel independent drive EV is relatively independent and singular, contradictions and conflicts inevitably arise when fulfilling corresponding control requirements. While traditional fault-tolerant algorithms offer strong reliability and robustness to handle actuator failures and restore system stability, no fault-tolerant algorithm has yet been proposed specifically for four-wheel independent drive EVs. Furthermore, fault-tolerant algorithms must be based on fault isolation and fault estimation to provide a solid foundation for the lower-level chassis controller. Especially with the addition of multiple onboard sensors, the accuracy of fault estimation also warrants further research. Therefore, four-wheel independent drive EVs also face the challenge of precise control of each subsystem, particularly when each subsystem malfunctions. Summary of the Invention

[0004] The purpose of this invention is to overcome the existing technical barriers and provide a fault diagnosis method (including fault isolation and fault estimation) for a four-wheel independent drive electric vehicle configuration. The proposed fault diagnosis method can not only meet the requirements of accurate fault coefficient estimation for electric vehicles under uncertain external interference and perturbation speed, but also make full use of the redundant configuration of four-wheel independent drive to accurately locate the faulty wheel hub motor, provide effective parameters for the chassis controller, and help improve the reliability and safety of the whole vehicle system.

[0005] The present invention solves the technical problem by adopting the following technical solution, including the following steps:

[0006] A method for diagnosing faults in the hub motors of a four-wheel independent drive electric vehicle, characterized by comprising:

[0007] A seven-degree-of-freedom vehicle dynamics equation was constructed, including longitudinal, lateral, and yaw plane motions as well as the motions of the four wheels. A hub motor fault model was established, including motor jamming faults and motor variable gain failure faults.

[0008] Based on the above model, the state space equation based on the Luenberger observer is established and the steady-state solution of the input and output information residuals is obtained. According to the plane projection theorem, by comparing the angle between the steady-state solution of the input residuals and the fault signal on the projected coordinate axis, the faulty hub motor is isolated.

[0009] Based on the above model and fault isolation signal, an H∞ robust fault estimator is built to accurately estimate the hub motor fault coefficient considering the uncertainty of vehicle speed perturbation and external disturbance.

[0010] By employing the above technical solutions, the present invention has the following beneficial effects compared to the prior art:

[0011] The invention fully utilizes the redundancy of four-wheel independent drive electric vehicles. Compared with the limitations of controllable units and the unreliability of detection algorithms in traditional electric vehicles, it reconstructs the vehicle dynamics model including hub motor faults, proposes a fault isolation mechanism based on mathematical model methods, and builds a robust fault estimator with strong fault tolerance and anti-interference capabilities. The fault diagnosis algorithm can greatly improve the fault tolerance and reliability of four-wheel independent drive electric vehicles in daily driving.

[0012] The advantages of this invention, such as good real-time performance and strong anti-interference ability, are beneficial for timely fault diagnosis in the actual driving process of four-wheel independent drive electric vehicles. It fills the technical gap that current sensor data cannot accurately estimate the fault coefficient and improves the applicability of this invention. Attached Figure Description

[0013] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0014] Figure 1 This is a fault isolation framework diagram in an example of the present invention.

[0015] Figure 2 This is a fault estimation framework diagram in an example of the present invention.

[0016] Figure 3 This is a flowchart of the system framework in an example of the present invention. Detailed Implementation

[0017] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention and therefore showing only the components relevant to the invention.

[0018] This invention proposes a fault diagnosis method for hub motors in four-wheel independently driven electric vehicles, effectively solving the problems of fault detection, fault isolation, and fault estimation in this novel configuration. Its framework diagram and flowchart are as follows: Figures 1-3 As shown, this invention proposes a hub motor fault isolation method based on the plane projection theorem, overcoming the shortcomings of existing methods such as model rigidity and poor real-time performance. A robust Ho fault estimator considering vehicle speed perturbations is constructed, minimizing fault characteristic signals and estimation coefficient residuals. This invention provides effective information for the chassis control layer, mitigating the risks of sensor and onboard component aging and circuit failures. The proposed fault detection, fault isolation, and fault estimation methods not only address vehicle speed perturbations and parameter uncertainties during actual vehicle operation but also significantly reduce external interference, exhibiting good accuracy and fault tolerance, thus improving the safety and stability of electric vehicles.

[0019] Step 1: Establish a dynamic model of a four-wheel independent drive electric vehicle, including longitudinal, lateral, and yaw motions in three planes, as well as the motions of the four vehicles. Key vehicle parameters can be obtained through real-vehicle verification, while real-time vehicle state variables are acquired from the CAN bus via onboard sensors or GPS, including: steering wheel angle signal δ. w Yaw velocity γ, wheel angular velocity ω i Longitudinal and lateral velocities V x and V y Assume the vehicle is traveling on a smooth road surface with a fixed coefficient of adhesion μ = 0.8. Considering the nonlinear characteristics of tire lateral stiffness under extreme conditions, the wheel lateral stiffness C... f and C r The nonlinearity of vehicle tire models can be solved by using a lookup table method to obtain the data in real time during experiments.

[0020] Step 11, the dynamic equations for the four-wheel independent drive electric vehicle in the longitudinal, lateral, and yaw planes are:

[0021]

[0022] Where m is the total mass of the vehicle. For longitudinal acceleration, For lateral acceleration, V x V is the longitudinal velocity. y γ is the lateral velocity, γ is the yaw rate, and F is the lateral velocity. x F y and Mz These are the longitudinal resultant force, lateral resultant force, and total yaw moment of the vehicle.

[0023] Step 12, assuming the front wheel steering angle δ f If the value of cosδ is small, then f ≈1, sinδ f ≈0, F is calculated x F y and M z They are respectively:

[0024]

[0025] The slip angles of the front and rear wheels can be expressed as:

[0026]

[0027] The lateral forces on each tire are:

[0028]

[0029] Among them, C f For the lateral stiffness of the front tire, C r α is the rear wheel tire lateral stiffness, μ is the road adhesion coefficient, and α is the road surface adhesion coefficient. sij F is the maximum tire slip angle when the tire lateral force is saturated. xfl F is the longitudinal force of the left front tire. xfr F is the longitudinal force on the right front tire. xrl F is the longitudinal force on the left rear tire. xrr F is the longitudinal force on the right rear tire. yfl F is the lateral force of the left front tire. yfr F is the lateral force of the right front tire. yrl F is the lateral force of the left rear tire. yrr The lateral force on the right rear tire;

[0030] Step 13, the four tire models are:

[0031]

[0032] in, Let T be the longitudinal angular acceleration of the i-th wheel tire. i The output torque of the hub motor for the i-th wheel is given by r, where r is the effective radius of the wheel. ω Let F be the moment of inertia of the wheel rotating about the y-axis. xi The driving force of the i-th wheel on the x-axis is given by , where i represents any one of the vehicle's left front wheel, right front wheel, left rear wheel, and right rear wheel.

[0033] Step 14: Establish the hub motor fault model. This invention assumes that only one hub motor fails at any given time, and that the fault types are motor jamming and variable gain failure. The fault model is as follows:

[0034]

[0035] Among them, T i.a Let k be the actual output torque value of the hub motor of the i-th wheel. i Let u be the fault coefficient of the i-th wheel hub motor. i The output torque is for the i-th wheel drive shaft. This is a deviation fault; when k i ∈[0,1] and Defined as a variable gain failure; when k i =0 and Defined as a motor jamming fault.

[0036] Step 15: Establish the vehicle state-space equations:

[0037]

[0038] Wherein, the state variable x = [γ V x ] T Control quantity u = [u fl u fr u rl u rr ] T Interference δ f and The problem is related to the front wheel steering angle and deviation.

[0039] The second step involves applying the obtained state-space equations to subsequent theoretical analysis, establishing a state observer, and deriving the steady-state solution from the input and output residual information. The planar projection method is then used to meet the fault isolation requirements, effectively detecting faulty hub motors in real time. The fault isolation framework diagram is shown below. Figure 1 As shown:

[0040] Step 21: Establish the state-space equations that take into account actuator fault signals using the Luenberger state observer:

[0041]

[0042] in, These are the state vector and the state estimation vector, respectively. These represent the measurement output and the measurement estimation output, respectively; A, B, C, and D are the system matrices for constructing the fault detection observer; and G is the fault diagnosis inverse gain matrix.

[0043] The state residual e and the output residual r are as follows:

[0044]

[0045] Step 22, combining the above fault detection filter, yields the state error equation and residual equation as follows:

[0046]

[0047] Step 23, consider the fault mathematical model under the condition that the i-th hub motor fails:

[0048] u = u d +ε j n(τ)

[0049] In the formula, u d ε is the ideal control input for the actuator under normal operating conditions. j ε represents the unit vector in the j-th coordinate direction. j =[0…010…0] T n(τ) represents the time scalar at time τ, where τ is the time it takes for the system to eventually reach stability.

[0050] Step 24, combining the mathematical model of hub motor fault and the state equation of the fault detector, is as follows:

[0051] X(t)=AX(t)+Bu d (t)+ε j n(t) + DW(t)

[0052] Step 25, when the hub motor fails, the residual equation is:

[0053]

[0054] Step 26: By analyzing the input and output residual equations, solutions for the two types of fault models can be obtained:

[0055]

[0056] The first term is the transient solution; the second term is the steady-state solution. Assuming the system remains stable, we only need to consider the steady-state solutions of the hub motor's input and output residual equations:

[0057]

[0058] Step 27: The fault detection and isolation logic of this invention mainly identifies and locates the faulty hub motor by calculating the input residuals. It compares the magnitude of the input residuals for each hub motor; however, determining the threshold value of the r-th input residual is very difficult, making effective detection and isolation of the faulty hub motor challenging. Therefore, this invention detects and separates faults by comparing the angle between the input residual e(t) and its projection, as well as the projection distance. The principle is to calculate the steady-state solution e(t) of the input residual for the i-th actuator that has failed. a If the angle between the input residual e and the fault feature vector is the largest, it proves that the input residual e is closest to the fault feature signal of the i-th actuator and that it is the hub motor most likely to fail.

[0059] Step 28: According to the plane projection theorem, the steady-state solution e(t) of the input residual when the i-th actuator fails is obtained. a The cosine of the angle between the fault eigenvector and the fault feature vector can be obtained using the inner product theorem:

[0060]

[0061] In the formula, CORR i π is the correlation coefficient of the fault isolation signal for the i-th wheel hub motor; i =λ j n(τ) is a time scalar function, λ j Let |j| represent the error signal of the j-th actuator; |·| represents the absolute value, and ||·|| represents the 2-norm of the vector.

[0062] Step 29: Calculate the relevant vector parameters of the four hub motors and sort them by size. The hub motor with the largest relevant vector parameters is the faulty hub motor that needs to be isolated.

[0063] ISO i =max[CORR fl CORR fr CORR rl CORR rr ]

[0064] In the formula, ISO i To definitively identify the isolated, failed hub motor, CORR fl CORR fr CORR rl and CORR rr These represent the correlation coefficients of the hub motor fault isolation signals for the left front wheel, right front wheel, left rear wheel, and right rear wheel, respectively.

[0065] Step 3: Based on the state-space equations and fault isolation information established above, a robust H∞ fault estimator is built, which minimizes the residual values ​​of fault characteristic signals and estimation coefficients, takes into account the uncertainties of vehicle speed perturbation and external interference, and significantly improves the feasibility of fault detection in actual driving processes.

[0066] Step 31: After isolating the faulty hub motor, the system state-space equation is established as follows:

[0067]

[0068] This paper uses polyhedral linear time-varying parameters to estimate the fault coefficients of each hub motor. By constructing a closed-loop system that considers fault signals, the estimated fault coefficients are made infinitely close to their actual values ​​to achieve real-time estimation. p (θ), B p (θ), C p (θ), D p (θ), E p (θ), F p (θ) is a real matrix considering parameter perturbation, f is the fault signal, and x p For system state variables, u p Let w be the system input, and w be the uncertain external disturbance. The derivative of the system state variables; y p For system quantity measurement.

[0069] Step 32: To design a robust fault estimator that considers the LPV model inequalities and minimizes the fault estimation error between the estimated and actual fault coefficients, the robust H∞ structure of the LPV estimator should utilize a feedback control method to establish the fault estimator E(s, θ):

[0070]

[0071] In the formula, x f The state of the fault estimator. This is an estimated value for the fault signal. Let A be the input vector. f (θ), B f (θ), C f (θ) and D f (θ) is a matrix to be determined.

[0072] Step 33: Combine the LPV model P(s, θ) and the fault estimation model E(s, θ) to obtain a new closed-loop system:

[0073]

[0074] In the formula,

[0075] A cl (θ)=A0(θ)+B k F(θ)C k B cl (θ)=B0(θ)+B k F(θ)D k21

[0076] C cl (θ)=C0(θ)+D k12 F(θ)C k D cl (θ)=D 11 (θ)+D k12 F(θ)D k21

[0077] Step 34, after obtaining the closed-loop value x cl Previously, some assumptions needed to be set:

[0078] Step 341, 1). The time-varying parameter θ(t) is at the vertices θ1, θ2, ... θ of the polyhedron Ξ. m Internal changes can be expressed as:

[0079]

[0080] The vertices in this equation reflect the range of values ​​for each parameter. Here, we can assume that parameter V... y and 1 / V x It is relatively independent, and its perturbation range varies within a certain range in each step. In this paper, considering the accuracy of the vehicle state information obtained from the vehicle sensors, the vehicle longitudinal velocity Vx should be set to a median or high value to ensure the feasibility of the proposed estimation algorithm.

[0081] Step 342, linear time-varying parameter V y and 1 / V x The envelope formed by these lines is constrained within a certain range.

[0082]

[0083]

[0084] Step 343, matrix C p (θ), D p (θ), E p (θ) and F p The parameters of (θ) are relatively independent and can be expressed as:

[0085]

[0086] Step 35, the goal of the H∞ fault estimator is to minimize the difference between the actual fault coefficient and the estimated fault coefficient: In addition, in order to reduce the impact of input w on system output, a feedback estimation controller E(s, θ) that satisfies closed-loop performance is designed based on the principle of robust asymptotic stability.

[0087] Step 36: To obtain the matrix of the closed-loop system, two conditions must be met:

[0088] Step 361, 1) For all possible parameter values ​​θ(t), the LPV system should be quadratic stable, and the infinite norm z induced by the closed-loop transfer function from the disturbance input d to the feedback output should be less than ε.

[0089] ||z||2<ε||w||2

[0090] ε represents the performance of the controller and should be as small as possible to reduce the impact of disturbances.

[0091] Step 362, 2). The linear matrix inequality has a positive definite symmetric solution x. cl >0.

[0092]

[0093] In the formula, / is a 5*5 identity matrix.

[0094] Step 37: To solve the LPV matrix inequality, the substitution method is used to transform X... cl Replacing the matrix with two given matrices R and S simplifies the system, the entire closed-loop system can be represented as:

[0095]

[0096]

[0097]

[0098] Step 38 It is a matrix [C2 D] 21 The null basis of [R, S] should satisfy the restriction 0 < R ∈ R. n×n , 0 < S ∈ S n×n And A m B 1m These are the values ​​of the parameter polyhedra A(θ) and B1(θ) at vertex θ.

[0099] Step 381: Prove that a sufficient prerequisite for the feasibility of the LPV fault estimation controller is the existence of a positive definite symmetric solution 0 < X. cl ∈R (n+k)×(n+k) The linear matrix inequality can be transformed into:

[0100]

[0101] In the formula:

[0102]

[0103] Q = [CD] 21 0 (k+p)×p ]

[0104] Step 382, ​​prove that the sufficient and preconditions for the feasibility of matrices R and S satisfying the linear matrix inequality are: matrix X CL It is positive definite and satisfies the inequalities mentioned earlier. X CL The value can be substituted into the equation to obtain the real-time value:

[0105] MN T =I-RS

[0106]

[0107] Step 39: Solve the linear matrix inequalities using convex optimization methods to obtain the matrix of the fault estimator. All of these linear matrix inequalities can be obtained through two iterations using the LMI toolbox.

[0108] Step 391: After acquiring the estimated fault signal, the output weight matrix is ​​adjusted accordingly to minimize the estimation error. The output weight matrix is ​​designed as follows:

[0109] Q e =diag(Q Af Q Bf Q Cf Q Df )

[0110] In the formula, Q Af Q Bf Q Cf Q Df Acting on system matrix A respectively cl B cl C cl D cl The weighting coefficients.

[0111] Step 392, the obtained full output matrix Q is... e and state variable X cl Substituting these values ​​into the fault estimator E(s, θ), the final hub motor fault coefficient can be obtained.

Claims

1. A wheel hub motor fault diagnosis method suitable for a four-wheel independent drive electric vehicle, characterized in that, Includes the following steps: A seven-degree-of-freedom vehicle dynamics model and a hub motor fault model were established for four-wheel independent drive electric vehicles. The seven degrees of freedom include longitudinal, lateral, and yaw motions in three planes, as well as the motions of the four wheels. The hub motor fault model includes variable gain failure and motor jamming failure. Based on the established seven-degree-of-freedom vehicle dynamics model and hub motor fault model, a fault detection state space equation based on a state observer is established to obtain the steady-state solutions of the input and output information residuals. According to the plane projection theorem, the steady-state solutions of the input residuals of the hub motor fault and the cosine values ​​of the angles between the fault signals and the projected coordinate axes are calculated. The cosine values ​​of the angles are used as the correlation coefficients of the fault isolation signals of the hub motor. The hub motor with the largest correlation coefficient is identified as the faulty hub motor, and fault isolation is performed on the hub motor with the fault. The formula for calculating the cosine of the included angle is: wherein is the fault isolation signal correlation coefficient of the i-th wheel hub motor; is a time scalar function, represents the error signal of the i-th hub motor; is the i-th hub motor; denotes the absolute value, denotes the 2-norm of a vector; After isolating the detected faulty hub motor, the system state-space equations are established, and a robust fault estimator considering LPV model inequalities is built to estimate the hub motor fault coefficients under uncertainties of vehicle speed perturbation and external disturbances. The established system state-space equations are as follows: In the formula, To account for the real matrix of vehicle speed perturbation, This is a fault signal. For system state variables, Here, w represents the system input, and w represents uncertain external disturbances. The derivative of the system state variables; For system quantity measurement.

2. The method for diagnosing hub motor faults in four-wheel independent drive electric vehicles according to claim 1, characterized in that: In the steps of establishing the seven-degree-of-freedom vehicle dynamics model and the hub motor fault model, the established seven-degree-of-freedom vehicle dynamics model is as follows: in, For the total mass of the vehicle. For longitudinal acceleration, For lateral acceleration, For longitudinal velocity, For lateral velocity, The yaw rate is angular velocity. The yaw acceleration is... wrap around the car body Moment of inertia of the axis of rotation These are the longitudinal resultant force, lateral resultant force, and total yaw moment of the vehicle, respectively. For the first The longitudinal angular acceleration of each wheel tire For the first The wheel hub motor outputs torque. The effective radius of the wheel, For the wheels Moment of inertia of the axis of rotation For the first One wheel in The driving force of the shaft, It represents any one of the vehicle's left front wheel, right front wheel, left rear wheel, and right rear wheel.

3. The method for diagnosing hub motor faults in four-wheel independent drive electric vehicles according to claim 2, characterized in that: In the steps of establishing a seven-degree-of-freedom vehicle dynamics model and a hub motor fault model applicable to four-wheel independent drive electric vehicles, the established hub motor fault model is as follows: in, For the first The actual output torque of the wheel hub motor for each wheel. For the first Failure coefficient of individual wheel hub motor For the first The output torque of each wheel drive shaft This is a deviation fault; when and Defined as a variable gain failure fault; when and Defined as a motor jamming fault.

4. The method for diagnosing hub motor faults in four-wheel independent drive electric vehicles according to claim 3, characterized in that: Based on the established seven-degree-of-freedom vehicle dynamics model and hub motor fault model, a fault detection state-space equation based on a state observer is established, and fault isolation is performed on the faulty hub motor according to the plane projection theorem, including: Establish the state-space equations using the Luenberger state observer: in, These are the state vector and the state estimation vector, respectively. These are the measurement output and the measurement estimation output, respectively. To construct the system matrix for the fault detection observer; This is the inverse gain matrix for fault diagnosis; The solution to the input-type fault model of the faulty hub motor is obtained based on the established state-space equations: In each formula, the first and second halves represent the transient and steady-state solutions of the input and output residuals, respectively. for The input residual solution and the output residual solution at time t; Representing the The unit vector in each coordinate direction, i.e. , represent The time scalar at a given moment. The time it takes for the system to eventually reach stability; Based on the obtained input residual solution and output residual solution, determine its steady-state solution: in, Let be the steady-state solution of the input residual of the i-th hub motor; Steady-state solution of the output residual of the i-th hub motor; Based on the plane projection theorem, calculate the steady-state solution of the input residual when the i-th hub motor fails. The cosine of the angle between the fault feature vector and the fault feature vector: In the formula, Let be the correlation coefficient of the fault isolation signal for the i-th wheel hub motor; It is a time scalar function. Representing the Error signals for each hub motor; Represents absolute value. The 2-norm of a vector; Calculate the correlation coefficients of the four hub motors. When a fault occurs, the hub motor with the highest correlation coefficient is the faulty hub motor. In the formula, To finally identify the isolated faulty hub motor; These represent the correlation coefficients of the hub motor fault isolation signals for the left front wheel, right front wheel, left rear wheel, and right rear wheel, respectively.

5. The method for diagnosing hub motor faults in four-wheel independent drive electric vehicles according to claim 4, characterized in that: Design a robust fault estimator that considers the LPV model inequality. Establish a closed-loop system that minimizes the fault estimation error between the estimated fault coefficient and the actual fault coefficient, and design a fault estimator using feedback control principles. for: In the formula, These are the substitution matrices for the fault estimator; This is an estimated value for the fault signal; These are the state variables and their derivatives of the fault estimator, respectively. LPV model and fault estimation model Combining them yields a new closed-loop system: In the formula, These are the state variables and their derivatives of the closed-loop system, respectively. , , and The system matrix that makes up the closed-loop system; For measurement output of the closed-loop system; The input of the closed-loop system and ; In the formula, and, In the formula, Let n be an identity matrix of dimension k; the subscript n*k represents an n-row k-column matrix; the subscript k*q represents a k-row q-column matrix; the subscript p*k represents a p-row k-column matrix. The values ​​are 2, 3, 4, and 5, representing the dimensions of the identity matrix; Obtain the vehicle longitudinal speed range considering external interference and communication delay. and lateral speed range It is necessary to establish the coordinates of the four vertices that take into account their perturbation range: In the formula, , ; For all possible parameter value ranges The LPV closed-loop system should be quadratic stable, and the infinite norm induced by the closed-loop transfer function from the disturbance input d to the feedback output should be constant. Less than : This represents the controller's ability to suppress external disturbances, satisfying the existence of a positive definite symmetric solution in the linear matrix inequality. ; In the formula, for The identity matrix; The linear matrix inequalities are solved using convex optimization methods to obtain the matrix of the fault estimator. ; matrix Substitute into the fault estimator In this process, based on the estimated error value of the hub motor fault coefficient, the offline output weight matrix is ​​manually modified and adjusted to obtain the isolated hub motor fault coefficient. The output weight matrix is: In the formula, Acting on the system matrix respectively The weighting coefficients.

Images