Robust fault-tolerant control method for automobile active suspension

Abstract

The invention discloses an automobile active suspension robust fault-tolerant control method, which belongs to the technical field of suspension control, and comprises the following steps: S1, carrying out disturbance observation on an active suspension system; s2, determining the fault level of the active suspension system after disturbance observation; and S3, performing robust fault-tolerant control on the suspension system according to the fault level of the active suspension system. For a system which still has an equivalent fault coefficient of 0.8 after fault-tolerant compensation, the robust fault-tolerant scheme designed by the invention can compensate the output of the system controller to 96% of the output of the system controller before the fault.

Description

Technical Field

[0001] This invention belongs to the field of suspension control technology, specifically relating to a robust fault-tolerant control method for active suspension in automobiles. Background Technology

[0002] As a crucial component of a vehicle, the suspension system significantly impacts overall ride comfort and handling stability. Traditional suspension systems, with their fixed parameters, struggle to adapt to all driving conditions. To provide a superior driving experience, suspension systems need to be able to adjust parameters in real time; hence, active suspension systems, which offer continuous active control, have become widely used. For active suspension systems, control algorithms are key to improving performance. A critical challenge lies in ensuring system robustness while enhancing suspension dynamics.

[0003] When a suspension system malfunctions, such as sensor failure or actuator failure, the vehicle's dynamic characteristics will change significantly, potentially leading to serious consequences such as rollover. Therefore, researching fault-tolerant control methods for suspension systems is of great importance. Fault-tolerant control systems can ensure the system can safely complete its control tasks according to its original performance targets or with slightly reduced performance targets even when certain components fail. In automotive suspension systems, active fault-tolerant control has become a research hotspot due to its fast response speed and strong adaptability.

[0004] On the one hand, many active suspension control methods have poor robustness and low reliability; on the other hand, most active suspension fault-tolerant control algorithms do not take into account the uncertainties of the suspension system itself. These uncertainties affect the control performance of the suspension system and also have a certain impact on the accuracy of fault detection and diagnosis of the suspension system. Summary of the Invention

[0005] To address the problem that most current suspension fault-tolerant control methods do not consider the system's own uncertainties and have poor robustness, this invention proposes a robust fault-tolerant control method for automotive active suspension.

[0006] The technical solution of this invention is: a robust fault-tolerant control method for automotive active suspension, comprising the following steps:

[0007] S1. Conduct disturbance observation on the active suspension system;

[0008] S2. Determine the fault level of the active suspension system after disturbance observation;

[0009] S3. Based on the fault level of the active suspension system, implement robust fault-tolerant control of the suspension system.

[0010] Furthermore, S1 includes the following sub-steps:

[0011] S11. Construct the state-space equations of the active suspension system;

[0012] S12. Use a disturbance observer to perform disturbance compensation on the active suspension system and obtain the state-space equation of the active suspension system after disturbance compensation.

[0013] S13. Based on the state-space equation of the active suspension system after disturbance compensation, the state of the active suspension system is observed using a Kalman filter to obtain the corresponding estimated values.

[0014] Furthermore, in S11, the state-space equation of the active suspension system is expressed as follows:

[0015] ;

[0016] in, The derivative of the system state variables. For the system's state variables, For the system output, For the system matrix, This is the first input matrix. This is the second input matrix. For the output matrix, It is a feedforward matrix. This is the second feedforward matrix. To further compensate for the output of the rear suspension actuator, This serves as the road surface excitation input for the system.

[0017] In S12, the expression for the state-space equation of the active suspension system after disturbance compensation is:

[0018] ;

[0019] in, The derivatives of the system state variables after disturbance compensation. These are the state variables of the system after disturbance compensation. This is the output of the system after disturbance compensation. The system matrix after disturbance compensation. This is the input matrix after perturbation compensation. This is the output matrix after disturbance compensation. This is the feedforward matrix after perturbation compensation. This is the first noise driving matrix. This is the second noise driving matrix. For the output force of the feedback controller, To measure the noise vector, This is the process noise vector;

[0020] In S13, the prediction equation of the Kalman filter is expressed as follows:

[0021] ;

[0022] ;

[0023] in, For based on Time information The predicted value of the state at time step. for State estimate at time 10:00 for The covariance matrix of the prediction error at each time step. Here is the state transition matrix. For the input matrix, The process noise input matrix, Let be the covariance matrix of the process noise. This is the transpose operation for a matrix;

[0024] In S13, the expression for the update equation of the Kalman filter is:

[0025] ;

[0026] ;

[0027] ;

[0028] in, For Kalman gain, for The covariance matrix of the time-state estimation error. To measure the noise covariance matrix, To measure the noise input matrix, For the observation matrix, For the input-observation matrix, for The state estimate after each update. for The predicted state value at time 10:00. for The observed value at time, for System input at any given time.

[0029] Furthermore, S2 includes the following sub-steps:

[0030] S21. Determine the actuator failure coefficient of the active suspension system after disturbance observation;

[0031] S22. Sample the residual information of suspension dynamic deflection under different actuator failure coefficients, and take the root mean square value of the sampling results as the residual threshold value for judging the failure level.

[0032] Furthermore, S3 includes the following sub-steps:

[0033] S31. Determine the additional compensation for the disturbance observer based on the fault level of the active suspension system;

[0034] S32. Determine the output of the suspension actuator after additional compensation;

[0035] S33. Based on the output of the rear suspension actuator after additional compensation, the improved robust LQR control method is used to compensate for the residual lumped disturbance in the active suspension system.

[0036] Furthermore, in S31, additional compensation for the disturbance observer. The expression is:

[0037] ;

[0038] in, This is the fault compensation coefficient. The failure factor is... For aggregated disturbances, To and Vectors of the same dimension and all elements being 1. The output force of the feedback controller;

[0039] In S32, the expression for the output of the additional compensation rear suspension actuator is:

[0040] ;

[0041] In S33, the improved robust LQR control method uses the output force of the feedback controller as the rLQR control input. The expression for the rLQR control input is:

[0042] ;

[0043] in, For rLQR control force equivalent term, For robust terms;

[0044] Furthermore, the rLQR control force equivalent term The expression is:

[0045] ;

[0046] in, To control the input weight matrix, The system input matrix after disturbance compensation. This is a solution to the Riccati equation. For controller state variables;

[0047] Robust item The expression is:

[0048] ;

[0049] in, A coefficient that satisfies certain conditions.

[0050] The beneficial effects of this invention are as follows: For systems that still have an equivalent fault coefficient of 0.8 after fault-tolerant compensation, the robust fault-tolerant scheme designed in this invention can compensate the system controller output to 96% of its pre-fault level. Compared to the faulty side of the active suspension without fault tolerance, the faulty side using fault-tolerant control does not experience deterioration in suspension dynamic deflection, and the tire dynamic travel is optimized by 50.0%, which is similar to the performance when no fault occurs, with a difference of only 1.3%. Attached Figure Description

[0051] Figure 1 A flowchart of a robust fault-tolerant control method for active suspension in automobiles;

[0052] Figure 2 The diagram shows the structure of a perturbation observer-based Kalman filter (DOB-KF).

[0053] Figure 3 This is a schematic diagram of a 4-DOF (degrees of freedom) half-vehicle active suspension model. Detailed Implementation

[0054] The embodiments of the present invention will be further described below with reference to the accompanying drawings.

[0055] like Figure 1 As shown, this invention provides a robust fault-tolerant control method for automotive active suspension, comprising the following steps:

[0056] S1. Conduct disturbance observation on the active suspension system;

[0057] S2. Determine the fault level of the active suspension system after disturbance observation;

[0058] S3. Based on the fault level of the active suspension system, implement robust fault-tolerant control of the suspension system.

[0059] The robust fault-tolerant control strategy of this invention is applicable to suspension systems with uncertainties. It ensures reliable compensation for actuator failures in the suspension system through a disturbance observer and the rLQR algorithm. The DOB-KF-based Kalman filter can eliminate the influence of slow time-varying disturbances and measurement noise that do not conform to the Gaussian white noise assumption, thus improving the accuracy and applicability of the Kalman filter.

[0060] like Figure 2 As shown, a Kalman filter (DOB-KF) based on a disturbance observer is designed for slow time-varying disturbances in the suspension system that do not satisfy the Gaussian white noise assumption. Based on this, a corresponding suspension control algorithm, fault diagnosis scheme and robust fault-tolerant control strategy are designed to optimize the performance and improve the robustness of the active suspension system.

[0061] In this embodiment of the invention, S1 includes the following sub-steps:

[0062] S11. Construct the state-space equations of the active suspension system;

[0063] S12. Use a disturbance observer to perform disturbance compensation on the active suspension system and obtain the state-space equation of the active suspension system after disturbance compensation.

[0064] S13. Based on the state-space equation of the active suspension system after disturbance compensation, the state of the active suspension system is observed using a Kalman filter to obtain the corresponding estimated values.

[0065] In this embodiment of the invention, in S11, the expression for the state-space equation of the active suspension system is:

[0066] ;

[0067] in, The derivative of the system state variables. For the system's state variables, For the system output, For the system matrix, This is the first input matrix. This is the second input matrix. For the output matrix, It is a feedforward matrix. This is the second feedforward matrix. To further compensate for the output of the rear suspension actuator, This serves as the road surface excitation input for the system.

[0068] In S12, the expression for the state-space equation of the active suspension system after disturbance compensation is:

[0069] ;

[0070] in, The derivatives of the system state variables after disturbance compensation. These are the state variables of the system after disturbance compensation. This is the output of the system after disturbance compensation. The system matrix after disturbance compensation. This is the input matrix after perturbation compensation. This is the output matrix after disturbance compensation. This is the feedforward matrix after perturbation compensation. This is the first noise driving matrix. This is the second noise driving matrix. For the output force of the feedback controller, To measure the noise vector, This is the process noise vector;

[0071] In S13, the prediction equation of the Kalman filter is expressed as follows:

[0072] ;

[0073] ;

[0074] in, For based on Time information The predicted value of the state at time step. for State estimate at time 10:00 for The covariance matrix of the prediction error at each time step. Here is the state transition matrix. For the input matrix, The process noise input matrix, Let be the covariance matrix of the process noise. This is the transpose operation for a matrix;

[0075] In S13, the expression for the update equation of the Kalman filter is:

[0076] ;

[0077] ;

[0078] ;

[0079] in, For Kalman gain, for The covariance matrix of the time-state estimation error. To measure the noise covariance matrix, To measure the noise input matrix, For the observation matrix, For the input-observation matrix, for The state estimate after each update. for The predicted state value at time 10:00. for The observed value at time, for System input at any given time.

[0080] In this embodiment of the invention, S2 includes the following sub-steps:

[0081] S21. Determine the actuator failure coefficient of the active suspension system after disturbance observation;

[0082] S22. Sample the residual information of suspension dynamic deflection under different actuator failure coefficients, and take the root mean square value of the sampling results as the residual threshold value for judging the failure level.

[0083] In this embodiment of the invention, S3 includes the following sub-steps:

[0084] S31. Determine the additional compensation for the disturbance observer based on the fault level of the active suspension system;

[0085] S32. Determine the output of the suspension actuator after additional compensation;

[0086] S33. Based on the output of the rear suspension actuator after additional compensation, the improved robust LQR control method is used to compensate for the residual lumped disturbance in the active suspension system.

[0087] In this embodiment of the invention, in S31, the additional compensation for the disturbance observer... The expression is:

[0088] ;

[0089] in, This is the fault compensation coefficient. The failure factor is... For aggregated disturbances, To and Vectors of the same dimension and all elements being 1. The output force of the feedback controller;

[0090] In S32, the expression for the output of the additional compensation rear suspension actuator is:

[0091] ;

[0092] The output force of the actuator serves as the control input to the system.

[0093] In S33, the improved robust LQR control method uses the output force of the feedback controller as the rLQR control input. The expression for the rLQR control input is:

[0094] ;

[0095] in, For rLQR control force equivalent term, For robust terms;

[0096] In this embodiment of the invention, the rLQR control force equivalent term The expression is:

[0097] ;

[0098] in, To control the input weight matrix, The system input matrix after disturbance compensation. This is a solution to the Riccati equation. For controller state variables;

[0099] Robust item The expression is:

[0100] ;

[0101] in, A coefficient that satisfies certain conditions.

[0102] In this embodiment of the invention, the state variables of the front and rear suspensions are read by sensors to construct a disturbance observation of the suspension system. This invention is based on a 4DOF half-vehicle suspension model considering pitch, and the motion differential equation of the active suspension system is:

[0103] ;

[0104] Rewriting the above equation as a state-space equation:

[0105] ;

[0106] In the formula, , , , , , , , , , , , , .

[0107] In the above equation, the system's state variables are: The system control input is The system's road surface excitation input is The system output is .

[0108] For simplicity, the uncertainty generated by the suspension system is considered as a lumped disturbance. and And rewrite the 4-DOF suspension model as follows:

[0109] ;

[0110] In the formula, Lumped disturbances of the front and rear suspensions The estimated value; This represents the gain of the perturbation observer.

[0111] The state-space equation of the system after disturbance compensation can be expressed as:

[0112] ;

[0113] In the formula, For the output force of the feedback controller, for A process noise vector of dimension, satisfying , and Lumped disturbance and The estimation error; for A dimensional measurement noise vector. It is assumed that both process noise and measurement noise conform to the Gaussian white noise assumption, i.e. Noise-driven matrix , They are as follows:

[0114] , .

[0115] The Kalman filter is designed as follows. Prediction equation:

[0116] ;

[0117] ;

[0118] Update equation:

[0119] ;

[0120] ;

[0121] ;

[0122] The disturbance and state of the suspension system are observed using the proposed DOB-KF (Disturbance Observer-based Kalman Filter) to obtain the corresponding estimates.

[0123] The active suspension actuator failure considered in this invention is caused by hydraulic oil leakage in the actuator. This type of failure is relatively common and changes slowly over time; therefore, it can be assumed that the actuator has experienced a constant gain failure. This invention only considers the case of a single suspension actuator experiencing a constant gain failure. For a 4-DOF active suspension system, the outputs of the front and rear suspension actuators after an actuator failure are as follows:

[0124] ;

[0125] In the formula, This represents the degree of failure of the suspension actuator. These represent the fault numbers of the front and rear active suspension actuators, respectively. At that time, the suspension did not fail. Define the actuator failure coefficient. .

[0126] The residual information of suspension dynamic deflection under fault coefficients of 0.1, 0.2, 0.3...0.9 was sampled, and the residual information for 200 seconds was obtained. The root mean square value of the residual information was taken as the residual threshold value for judging the current fault level. The fault levels and their corresponding coefficients are shown in Table 1.

[0127] Table 1

[0128] Actuator Fault Level Corresponding actuator failure coefficient 1 2 3 4 5 6 7 8 9 10

[0129] Robust fault-tolerant control of the suspension system consists of two parts:

[0130] 1) The DOB module can almost completely compensate for the fault compensation errors that may be introduced by the Fault Diagnosis Unit (FDD). As analyzed above, DOB has a certain fault tolerance capability; when the fault compensation coefficient is... After multiplicative fault tolerance compensation, assuming the system still exists The equivalent fault coefficient, which the disturbance observer can compensate for additionally:

[0131] ;

[0132] The output of the DOB-compensated suspension actuator is:

[0133] ;

[0134] 2) The improved robust LQR control method (rLQR) can compensate for residual lumped disturbances in the system while balancing multiple control objectives of the suspension system.

[0135] The system equations of the controller reference model are:

[0136] ;

[0137] In the formula, the controller state quantity Selected as , Residual lumped disturbance equivalent to the control channel ,satisfy .

[0138] The rLQR control input is designed as follows:

[0139] ;

[0140] In the formula, For rLQR control force equivalent term, For robustness terms. The performance functional is designed as follows:

[0141] ;

[0142] The equivalent term can then be expressed as:

[0143] ;

[0144] In the formula, The solution to the Riccati equation satisfies:

[0145] ;

[0146] Robust item satisfy:

[0147] ;

[0148] In the formula, Always greater than 0, and satisfying .

[0149] like Figure 3 As shown, This represents the vertical displacement of the vehicle body; The vertical displacement of the spring-loaded mass on the front / rear axle; The vertical displacement of the unsprung mass on the front / rear axle; The vehicle body pitch angle; For front / rear axle active suspension control force; The road surface excitation is for the front / rear axle. Table 2 shows the symbols and meanings of the parameters for the active suspension half-vehicle dynamics model.

[0150] Table 2

[0151] Symbols / Units describe Body sprung mass Front axle unsprung mass Rear axle unsprung mass Vehicle pitch moment of inertia Front axle suspension stiffness Rear axle suspension stiffness Front axle tire stiffness Rear axle tire stiffness Front axle shock absorber passive damping coefficient Rear axle shock absorber passive damping coefficient Distance from front axle to vehicle center of gravity Distance from rear axle to vehicle center of gravity wheelbase

[0152] Those skilled in the art will recognize that the embodiments described herein are intended to help the reader understand the principles of the invention, and should be understood that the scope of protection of the invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations based on the technical teachings disclosed in this invention without departing from the spirit of the invention, and these modifications and combinations are still within the scope of protection of this invention.

Claims

1. A robust fault-tolerant control method for active suspension in automobiles, characterized in that, Includes the following steps: S1. Conduct disturbance observation on the active suspension system; S2. Determine the fault level of the active suspension system after disturbance observation; S3. Based on the fault level of the active suspension system, implement robust fault-tolerant control of the suspension system.

2. The robust fault-tolerant control method for active suspension of an automobile according to claim 1, characterized in that, S1 includes the following sub-steps: S11. Construct the state-space equations of the active suspension system; S12. Use a disturbance observer to perform disturbance compensation on the active suspension system and obtain the state-space equation of the active suspension system after disturbance compensation. S13. Based on the state-space equation of the active suspension system after disturbance compensation, the state of the active suspension system is observed using a Kalman filter to obtain the corresponding estimated values.

3. The robust fault-tolerant control method for active suspension of an automobile according to claim 1, characterized in that, In S11, the state-space equation of the active suspension system is expressed as follows: ； in, The derivative of the system state variables. For the system's state variables, For the system output, For the system matrix, This is the first input matrix. This is the second input matrix. For the output matrix, It is a feedforward matrix. This is the second feedforward matrix. To further compensate for the output of the rear suspension actuator, This serves as the road surface excitation input for the system. In step S12, the state-space equation of the active suspension system after disturbance compensation is expressed as follows: ； in, The derivatives of the system state variables after disturbance compensation. These are the state variables of the system after disturbance compensation. This is the output of the system after disturbance compensation. The system matrix after disturbance compensation. This is the input matrix after perturbation compensation. This is the output matrix after disturbance compensation. This is the feedforward matrix after perturbation compensation. This is the first noise driving matrix. This is the second noise driving matrix. For the output force of the feedback controller, To measure the noise vector, This is the process noise vector; In S13, the prediction equation of the Kalman filter is expressed as follows: ； ； in, Based on Time information The predicted value of the state at time step. for State estimate at time 10:00 for The covariance matrix of the prediction error at each time step. Here is the state transition matrix. For the input matrix, The process noise input matrix, Let be the covariance matrix of the process noise. This is the transpose operation for a matrix; In S13, the expression for the update equation of the Kalman filter is: ； ； ； in, For Kalman gain, for The covariance matrix of the time-state estimation error. To measure the noise covariance matrix, To measure the noise input matrix, For the observation matrix, For the input-observation matrix, for The state estimate after each update. for The predicted state value at time 10:

00. for The observed value at time, for System input at any given time.

4. The robust fault-tolerant control method for active suspension of an automobile according to claim 1, characterized in that, S2 includes the following sub-steps: S21. Determine the actuator failure coefficient of the active suspension system after disturbance observation; S22. Sample the residual information of suspension dynamic deflection under different actuator failure coefficients, and take the root mean square value of the sampling results as the residual threshold value for judging the failure level.

5. The robust fault-tolerant control method for active suspension of an automobile according to claim 1, characterized in that, S3 includes the following sub-steps: S31. Determine the additional compensation for the disturbance observer based on the fault level of the active suspension system; S32. Determine the output of the suspension actuator after additional compensation; S33. Based on the output of the rear suspension actuator after additional compensation, the improved robust LQR control method is used to compensate for the residual lumped disturbance in the active suspension system.

6. The robust fault-tolerant control method for active suspension of an automobile according to claim 1, characterized in that, In S31, the additional compensation for the disturbance observer The expression is: ； in, This is the fault compensation coefficient. The failure coefficient, For aggregated disturbances, To and Vectors of the same dimension and all elements being 1. The output force of the feedback controller; In S32, the expression for the output of the additional compensation rear suspension actuator is: ； In S33, the improved robust LQR control method uses the output force of the feedback controller as the rLQR control input, and the expression for the rLQR control input is: ； in, For rLQR control force equivalent term, It is a robust term.

7. The robust fault-tolerant control method for active suspension of an automobile according to claim 1, characterized in that, The rLQR control force equivalent term The expression is: ； in, To control the input weight matrix, The system input matrix after disturbance compensation. This is a solution to the Riccati equation. For controller state variables; The robustness term The expression is: ； in, A coefficient that satisfies certain conditions.

Images