Sensor fault-tolerant control method for steer-by-wire vehicle based on steering motor torque regulation

Abstract

A kind of sensor fault-tolerant control method of steer-by-wire vehicle based on steering motor torque regulation, comprising: establishing the dynamics model of steer-by-wire system;Construct integrated disturbance, and establish the state space model of steer-by-wire system;Construct supercoil sliding mode observer, real-time estimation is carried out to the state variable of steer-by-wire system and integrated disturbance using supercoil sliding mode error compensation algorithm, and the estimated angle and disturbance estimated value of observer are output;Form multi-source angle data;Calculate observer transient residual error, auxiliary transient residual error, and generate sensor soft isolation weight;Threshold determination is carried out to minimum steady-state residual error using schmidt hysteresis discriminator, and sensor fault flag is updated;Generation fusion angle value is generated according to preset fusion proportion, and fault-tolerant angle value is generated;Sliding mode fault-tolerant control law is constructed, and then steering motor control torque is calculated and output.The present application can improve the fault-tolerant control ability of modeling and sensor under complex disturbance.

Description

Technical Field

[0001] This invention relates to the field of intelligent vehicle steer-by-wire control technology, and more specifically to a sensor-tolerant control method for steer-by-wire vehicles based on steering motor torque adjustment. Background Technology

[0002] With the rapid development of intelligent vehicles, steer-by-wire (SBW) systems have become the core steering execution architecture for next-generation intelligent vehicles due to their ease of integration with chassis control. This architecture eliminates the mechanical connection between the steering wheel and the front wheels, providing a solid hardware foundation for active steering and intelligent control.

[0003] However, the demechanization of the structure also brings new challenges. The system is highly dependent on electrical signal transmission, and if the sensors malfunction or fail, it directly affects the accuracy of the state feedback, further compromising the stability of the steering system and even endangering vehicle driving safety. Therefore, how to improve the modeling and fault-tolerant control of sensors under complex disturbances is a technical problem that needs to be solved by those skilled in the art. Summary of the Invention

[0004] In view of this, the present invention provides a fault-tolerant control method for sensors of drive-by-wire vehicles based on steering motor torque regulation, so as to improve the modeling and fault-tolerant control capability of sensors under complex disturbances.

[0005] A sensor-tolerant control method for steerable vehicles based on steering motor torque regulation, comprising: Step S1: Based on the vehicle's two-degree-of-freedom dynamic equations, the steering motor's dynamic equations, the front wheel's dynamic equations, and the transmission relationship, establish a dynamic model of the steer-by-wire system; Step S2: Based on the dynamic model of the steer-by-wire system established in Step S1, the nonlinear friction of the actuator, gear meshing error, road load change, low-frequency time-varying drift, high-frequency harmonic fluctuation and random environmental noise are equivalent to integrated disturbances, and a state space model of the steer-by-wire system is established. Step S3: Based on the state space model of the steer-by-wire system established in Step S2, construct a super-spiral sliding mode observer, define the state estimation error and the integrated disturbance estimation residual of the super-spiral sliding mode observer, and thus construct the error state equation. Based on the error state equation, use the super-spiral sliding mode error compensation algorithm to estimate the state variables and integrated disturbance of the steer-by-wire system in real time, so that the state estimation error and the integrated disturbance estimation residual converge in finite time, and finally output the observer's estimated steering angle and disturbance estimation value. Step S4: Obtain the actual rotation angle measured by the sensor, and combine it with the estimated rotation angle obtained in step S3, as well as the kinematic auxiliary rotation angle calculated based on the vehicle's yaw rate, to form multi-source rotation angle data; Step S5: Using the multi-source rotation angle data, calculate the observer transient residual between the sensor measured rotation angle and the observer estimated rotation angle, and the auxiliary transient residual between the sensor measured rotation angle and the kinematic auxiliary rotation angle, respectively. Extract the minimum transient residual using the minimum value operator, and then use the Logistic function to perform nonlinear mapping on the minimum transient residual to generate sensor soft isolation weights. Step S6: Filter the transient residual of the observer and the auxiliary transient residual using a first-order discrete low-pass filter to obtain the steady-state residual of the observer and the auxiliary steady-state residual. Then, extract the minimum steady-state residual using a minimum value operator. Finally, use a Schmitt hysteresis discriminator to determine the threshold of the minimum steady-state residual and update the sensor fault flag bit. Step S7: Based on the sensor fault flag bit updated in step S6, the observer estimated rotation angle and the kinematic auxiliary rotation angle, generate a fused rotation angle value according to a preset fusion ratio, and then combine the sensor soft isolation weight generated in step S5 with the sensor measured rotation angle to perform weighted reconstruction and generate a fault-tolerant rotation angle value. Step S8: Reconstruct the system state variables using the fault-tolerant angle value, construct a sliding mode fault-tolerant control law based on the reconstructed state variables, and finally calculate and output the steering motor control torque through the sliding mode fault-tolerant control law.

[0006] The fault-tolerant control method for drive-by-wire vehicles based on steering motor torque regulation provided by the present invention has the following advantages: (1) The present invention constructs a dynamic model of SBW system containing complex nonlinear disturbances. By equating the nonlinear friction of the actuator, gear meshing error, road load change, low-frequency time-varying drift, high-frequency harmonic fluctuation and random environmental noise as integrated disturbance terms and incorporating them into the system state space model, the model's representation accuracy of the real working conditions is effectively improved, providing a more accurate model basis for subsequent observer design and fault-tolerant control.

[0007] (2) This invention designs a superspiral sliding mode observer, which effectively solves the problem of state estimation under complex disturbances. The observer uses unmodeled states such as nonlinear friction as integrated disturbance terms for real-time estimation and compensation. By utilizing the finite-time convergence characteristics of the superspiral sliding mode error compensation algorithm, it achieves high-precision real-time estimation of system state variables and equivalent integrated disturbances, thereby improving the state estimation accuracy and robustness of the SBW system.

[0008] (3) This invention proposes a fault-tolerant control architecture based on multi-source data fusion. This architecture constructs a pre-soft isolation and exponential smooth transition mechanism. That is, it generates soft isolation weights through the Logistic function of the transient fast channel to achieve fast response, and achieves false alarm avoidance through the Schmitt hysteresis discriminator of the steady-state slow channel. In the signal reconstruction, a fusion ratio of 9:1 is adopted, which can achieve smooth reconstruction of the steering angle signal under complex disturbance and sensor failure conditions, reduce the step impact of steering motor torque caused by diagnostic delay, false alarm and hard switching, and improve the tracking stability, fault tolerance safety and smoothness of the steer-by-wire system. Attached Figure Description

[0009] Figure 1 This is a flowchart illustrating the sensor-tolerant control method for drive-by-wire vehicles based on steering motor torque regulation, as provided in an embodiment of the present invention. Figure 2 This is a comparison chart of the observer's performance under the dual-track shifting condition; Figure 3 This is a comparison chart of tracking performance under dual-track shifting conditions; Figure 4 This is a comparison chart of tracking error and output torque under dual lane change conditions; Figure 5 A comparison chart of the observer's performance under fishhook operating conditions; Figure 6 Comparison of tracking performance under fishing hook working conditions; Figure 7 This is a comparison chart of tracking error and output torque under fishing hook operating conditions. Detailed Implementation

[0010] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain embodiments of the present invention, and should not be construed as limiting the present invention.

[0011] Please see Figure 1 The present invention provides a fault-tolerant control method for steerable vehicle sensors based on steering motor torque regulation, the method comprising steps S1 to S8: Step S1: Based on the vehicle's two-degree-of-freedom dynamic equations, the steering motor's dynamic equations, the front wheel's dynamic equations, and the transmission relationship, establish a dynamic model of the steer-by-wire system.

[0012] The process of establishing the dynamic model of the steer-by-wire system is a conventional method and will not be elaborated here.

[0013] Step S2: Based on the dynamic model of the steer-by-wire system established in Step S1, the nonlinear friction of the actuator, gear meshing error, road load change, low-frequency time-varying drift, high-frequency harmonic fluctuation and random environmental noise are equivalent to integrated disturbances, and a state space model of the steer-by-wire system is established.

[0014] In the process of equating the nonlinear friction of the actuator, gear meshing error, road load variation, low-frequency time-varying drift, high-frequency harmonic fluctuations, and random environmental noise as an integrated disturbance, the following equation is satisfied.

[0015]

[0016] in, Indicates integrated disturbance. Represents the Coulomb friction coefficient. Represents a symbolic function. The angular velocity of the front wheel's steering angle. It is a composite perturbation function. For low-frequency disturbance coefficients, For high-frequency disturbance coefficients, This represents the environmental noise disturbance factor. This is the low-frequency disturbance frequency. For high-frequency disturbance frequencies, For random environmental noise, For time.

[0017] Specifically, the first item Used to characterize the Coulomb friction effect caused by nonlinear friction in the actuator; composite disturbance function Low-frequency components Used to characterize road load changes and low-frequency time-varying drift, high-frequency components Used to characterize gear meshing error and high-frequency harmonic fluctuations, random term Used to characterize environmental noise.

[0018] Considering external disturbances to the vehicle, and taking the front wheel steering angle, front wheel steering angular velocity, vehicle center of gravity sideslip angle, and yaw rate as system state variables, a state-space model of the steer-by-wire system reflecting the vehicle's physical dynamics is established, expressed as:

[0019]

[0020]

[0021] in, Indicates the front wheel steering angle. Angular acceleration representing the steering angle of the front wheel. This represents the sum of the front wheel aerodynamic trail and the mechanical trail. For the front wheel lateral stiffness, For rear wheel lateral stiffness, For the equivalent front wheel rotational inertia, The equivalent friction coefficient, The sideslip angle is the angle of the center of mass. The angular velocity of the center of mass deflection. This is the distance from the front axle to the center of gravity. This is the distance from the rear axle to the center of gravity. For the longitudinal speed of the vehicle, The yaw rate is angular velocity. This is the yaw acceleration. The steering gear ratio, For the steering motor to output torque, For vehicle quality, This is the moment of inertia of the entire vehicle about the vertical direction.

[0022] Step S3: Based on the state space model of the steer-by-wire system established in step S2, construct a super-spiral sliding mode observer, define the state estimation error and the integrated disturbance estimation residual of the super-spiral sliding mode observer, and thus construct the error state equation. Based on the error state equation, use the super-spiral sliding mode error compensation algorithm to estimate the state variables and integrated disturbance of the steer-by-wire system in real time, and finally output the observer's estimated steering angle and disturbance estimate.

[0023] The superspiral sliding mode observer constructs sliding mode variables based on the system state estimation error and estimates the integrated disturbance through the superspiral sliding mode error compensation algorithm, so that the state estimation error and the integrated disturbance estimation residual converge in finite time.

[0024] Specifically, the superhelical sliding mode observer satisfies the following equation:

[0025]

[0026] in, These are estimates of the state variables. for The first derivative, To provide an estimate of the output variable, For integrated disturbance The estimated value, For sliding mode variables, For the system control law, The observer feedback gain matrix, , , , This is the system matrix.

[0027] Among them, matrix Satisfying Hurwitz stability, Indicates the selection vector. , Table transpose.

[0028] The expression for the system matrix is:

[0029]

[0030]

[0031] .

[0032] The state estimation error and the integrated perturbation estimation residual of the superspiral sliding mode observer are defined as follows:

[0033]

[0034] in, For state estimation error, To integrate the disturbance estimation residuals, For state variables, This is the error in the front wheel steering angle estimation. The angular velocity estimation error for the front wheel steering angle. This is the error in estimating the centroid sideslip angle. This represents the estimation error of the yaw rate.

[0035] Since the front wheel steering rate can be obtained through sensors, the observer in this invention uses the estimation error of the second state variable as the sliding surface feedback quantity, i.e. .

[0036] In step S3, combining the state-space model of the steer-by-wire system and the super-helical sliding mode observer, the error state equation is constructed, and its expression is:

[0037] in, yes The first derivative; To facilitate the subsequent calculation of the superspiral algorithm, an equivalent perturbation is defined. Estimated value of equivalent disturbance Equivalent disturbance error ; To ensure accurate estimation of the disturbance term by the observer within a finite time period, the superspiral sliding mode error compensation algorithm used satisfies the following equation:

[0038]

[0039] in, , For observer gain parameters, , , For observer intermediate variables, It is the first derivative of the intermediate variable of the observer.

[0040] Step S4: Obtain the actual rotation angle measured by the sensor, and combine it with the estimated rotation angle obtained in step S3, as well as the kinematic auxiliary rotation angle calculated based on the vehicle's yaw rate, to form multi-source rotation angle data.

[0041] Step S5: Using the multi-source rotation angle data, calculate the observer transient residual between the sensor measured rotation angle and the observer estimated rotation angle, and the auxiliary transient residual between the sensor measured rotation angle and the kinematic auxiliary rotation angle. Extract the minimum transient residual using the minimum value operator, and then use the Logistic function to perform nonlinear mapping on the minimum transient residual to generate sensor soft isolation weights.

[0042] In this embodiment, a transient fast channel is constructed based on the observer transient residual and the auxiliary transient residual. The core task of the transient fast channel is to react quickly to small changes in potential faults.

[0043] In step S5, the observer transient residual and the auxiliary transient residual are calculated using the following formula:

[0044]

[0045] in, For the transient residual of the observer, To assist in transient residuals, The actual rotation angle measured by the sensor. Estimate the rotation angle for the observer. To assist in kinematic rotation; To ensure that no potential risks are overlooked under any unknown nonlinear perturbation, this invention employs a minimum value operator to extract the minimum transient residual, expressed as:

[0046] in, It represents the minimum transient residual.

[0047] To address minor glitches or early sensor drift faults, the system abandons the traditional hard-switching logic that easily induces torque shocks, and introduces a Logistic function to calculate the sensor soft isolation weights. To prevent unnecessary weight attenuation caused by normal road surface white noise, an absolute dead zone is set in the system. , Tolerance warning threshold, in order to extract effective residuals When the error exceeds the dead zone and approaches the tolerance threshold, the weights exhibit an S-shaped smooth drop, achieving a non-linear smooth mapping between the error and the weights.

[0048] Specifically, the calculation process for the sensor soft isolation weight is as follows:

[0049]

[0050] in, For effective residuals, For sensor soft isolation weights, The shape factor of the Logistic function that determines the isolation rate. It is a natural exponential function.

[0051] In this invention, a dual fault-tolerant architecture is formed by a transient fast channel and a steady-state slow channel. The above process can reduce the weight of suspicious signals without triggering a formal fault alarm.

[0052] Step S6: Filter the transient residual of the observer and the auxiliary transient residual using a first-order discrete low-pass filter to obtain the steady-state residual of the observer and the auxiliary steady-state residual. Then, extract the minimum steady-state residual using a minimum value operator. Finally, use a Schmitt hysteresis discriminator to determine the threshold of the minimum steady-state residual and update the sensor fault flag bit.

[0053] While transient extremum evaluation has an extremely fast response, it is highly susceptible to false alarms caused by strong transient interference. To address this, this invention constructs a steady-state slow channel, in which a first-order discrete low-pass filter is introduced to independently extract the pure steady-state characteristics of the two signals.

[0054] Specifically, a first-order discrete low-pass filter is used to filter the observer transient residual and the auxiliary transient residual respectively, as expressed by:

[0055]

[0056]

[0057] in, These are the filter coefficients. The time constant of the low-pass filter. The sampling period is For discrete sampling time indexes, for The steady-state residual of the observer at time t. for The auxiliary steady-state residual at each moment, for The steady-state residual of the observer at time t. for The auxiliary steady-state residual at each moment, for The transient residual of the observer at time t, for Auxiliary transient residuals at time.

[0058] To maximize robustness against disturbances and reduce false alarm rate, the steady-state diagnostic layer is switched to the minimum operator, meaning that the diagnostic process only begins when both the state observer and the analytical model identify a steady-state deviation.

[0059] Specifically, the minimum steady-state residual is calculated using the minimum value operator, and the expression is:

[0060] in, To minimize the steady-state residual, For the steady-state residuals of the observer, To assist in steady-state residuals.

[0061] To eliminate the flickering phenomenon caused by residual fluctuations around the diagnostic threshold, a Schmitt Trigger is used in the diagnostic process to update the sensor fault flag, with the following expression:

[0062] in, The threshold for diagnosis for The sensor fault flag bit at any given time. for Sensor fault flag bit at any time, sensor fault flag bit This indicates that the sensor has malfunctioned. This indicates that the sensor is working properly.

[0063] Step S7: Based on the sensor fault flag bit updated in step S6, the estimated rotation angle of the observer and the kinematic auxiliary rotation angle, generate a fused rotation angle value according to a preset fusion ratio, and then combine the sensor soft isolation weight generated in step S5 with the measured rotation angle of the sensor to perform weighted reconstruction to generate a fault-tolerant rotation angle value.

[0064] To avoid the risk of long-term divergence caused by relying solely on the observer, the reconstruction strategy in this embodiment adopts a fusion ratio of 9:1, which integrates the steady-state characteristics of the observer with the dynamic characteristics of the auxiliary analytical equation.

[0065] Specifically, the tolerance angle value is calculated using the following formula:

[0066]

[0067] in, To integrate corner values, This is the fault-tolerant rotation angle value.

[0068] This invention, through the coordinated operation of fast and slow channels, not only ensures microsecond-level soft isolation in the face of sudden failures, but also avoids false alarms through rigorous steady-state dual diagnostics, eliminating the torque step phenomenon of the actuator motor.

[0069] Step S8: Reconstruct the system state variables using the fault-tolerant angle value, construct a sliding mode fault-tolerant control law based on the reconstructed state variables, and finally calculate and output the steering motor control torque through the sliding mode fault-tolerant control law.

[0070] Specifically, step S8 includes: The fault-tolerant rotation angle value generated in step S7 Using reliable front wheel steering angle information at the current moment as a reference, and combining it with the estimated state variables (including estimated values ​​of front wheel steering angle angular velocity, center of mass sideslip angle, and yaw rate) estimated by the super-spiral sliding mode observer, the reconstructed state variables are constructed. .

[0071] To ensure that the system maintains excellent trajectory tracking performance after signal reconstruction, a feedback tracking error is defined. The expression is:

[0072] in, This is the desired reference state; Redesign the sliding surface The expression is:

[0073] in, Here is the sliding surface parameter matrix; To mitigate the chattering effect caused by high-frequency switching and to enable the system state to converge rapidly to the sliding surface, this invention constructs an exponential reaching law using a saturation function. The expression is:

[0074] in, For the gain of the switching item, , For exponentially approaching gain, , It is a saturation function; Saturation function The definition of is:

[0075] in, This is the input quantity for the saturation function; Finally, a sliding mode fault-tolerant control law is constructed. The expression is:

[0076] in, Represents the rate of change of the desired reference state, and is a sliding mode fault-tolerant control law. The output is the control torque for the steering motor.

[0077] The following simulation test is conducted on the present invention. The test conditions selected for the present invention are the double line shifting condition and the fishhook condition, and the Extended Kalman (EKF) algorithm is used as the comparison algorithm.

[0078] Figure 2 The figure shows a performance comparison of the observer under the double-track shifting condition, where (a) represents the estimated rotation angle of the present invention. and EKF estimated turning angle With real corner (a) is a comparison chart; (b) is the disturbance estimate of the present invention. Compared with the disturbance reference value (c) shows the comparison chart; (d) shows the error between the estimated value of this invention and the actual rotation angle. Curve graph; (d) shows the estimated angular velocity of this invention. and EKF estimated angular velocity Compared with the actual angular velocity The comparison chart.

[0079] Please see Figure 2 Analysis of the observer's performance during the dynamic alternation period (4s-8s) reveals that the Extended Kalman Filter (EKF) algorithm forcibly truncates higher-order dynamic terms when linearizing the Jacobian matrix of nonlinear friction. This results in significant phase delay and amplitude attenuation in the estimated angle and angular velocity at peaks and troughs. In contrast, this invention, leveraging the finite-time convergence characteristic of the superspiral compensation mechanism, not only achieves hysteresis-free, high-fidelity tracking of the real state but also accurately reconstructs the comprehensive nonlinear disturbances accompanying the high-frequency oscillations of the steering motion, thus providing an extremely reliable data foundation for the subsequent dimensionality reduction and reconstruction of the controller.

[0080] Figure 3 The figures show a comparison of tracking performance under dual-track shifting conditions, where (a) represents the reconstructed turning angle of the present invention. With sensor output angle EKF Reconstructed Corner Reference angle (a) Comparison chart; (b) Observer transient residual With auxiliary transient residual Minimum transient residual (c) shows the comparison diagram; (c) shows the steady-state judgment process in the fault-tolerant architecture, and the steady-state residual of the observer. With the transient residual of the observer Minimum steady-state residual The comparison diagram shows the sensor fault flag on the right axis; (d) represents the error generated by the dual fault-tolerant architecture (i.e., transient fast channel and steady-state slow channel) and the reference signal in this invention. The error generated by the traditional hard-switching architecture (which directly uses the observer's estimated rotation angle plus the rotation angle calculated by the auxiliary equation, and outputs the average after summing, without filtering) compared to the reference signal The comparison chart.

[0081] After introducing a sensor malfunction Figure 3 This invention demonstrates a two-layer fault-tolerant architecture with a fault-tolerant mechanism for multi-source data. Addressing the critical issue of high-frequency road noise easily causing critical misjudgments, this invention employs a dual diagnostic approach combining low-pass filtering and a Schmitt hysteresis discriminator to fundamentally eliminate the flickering and oscillation of fault flag bits common in conventional logic. During the transient switching period after diagnosis, traditional EKF combined with a hard switching mechanism inevitably induces a step jump in the reconstructed signal; however, this invention, relying on a pre-positioned soft isolation channel based on Logistic attenuation characteristics, effectively reduces the error shear slope, achieving a smooth transition of the fault-tolerant feedback signal.

[0082] Figure 4 This is a comparison chart of tracking error and output torque under dual lane change conditions, where (a) is the fault-tolerant rotation angle of the present invention. EKF fault-tolerant rotation angle (a) Comparison chart; (b) Output torque of the present invention With EKF output torque The comparison chart.

[0083] Figure 4 This study reveals the substantial impact of signal-level isolation mechanisms on the closed-loop performance of underlying actuators. The combined effect of EKF estimation lag and traditional hard handover strategies directly induces… Figure 4 (b) describes the oscillation of motor torque. Such sudden torque changes can easily worsen the driver's road feel and exacerbate mechanical friction. In contrast, this invention, through the smooth fusion of multi-source data, eliminates the excitation signal that causes sudden torque changes at the source. Its output motor torque curve maintains excellent continuity and smoothness, achieving a safe switch from normal mode to fault-tolerant mode.

[0084] Figure 5 The graph shows a performance comparison of the observer under the fishing hook condition, where (a) represents the estimated rotation angle of the present invention. and EKF estimated turning angle With real corner (a) is a comparison chart; (b) is the disturbance estimate of the present invention. Compared with the disturbance reference value (c) shows the comparison chart; (d) shows the error between the estimated value of this invention and the actual rotation angle. Curve graph; (d) shows the estimated angular velocity of this invention. and EKF estimated angular velocity Compared with the actual angular velocity The comparison chart.

[0085] Figure 6 The images show a comparison of tracking performance under fishing hook conditions, where (a) shows the reconstructed turning angle of the present invention. With sensor output angle EKF Reconstructed Corner Reference angle (a) Comparison chart; (b) Observer transient residual With auxiliary transient residual Minimum transient residual (c) shows the comparison diagram; (c) shows the steady-state judgment process in the fault-tolerant architecture, and the steady-state residual of the observer. With the transient residual of the observer Minimum steady-state residual The comparison diagram shows the sensor fault flag on the right axis; (d) represents the error generated by the dual fault-tolerant architecture (i.e., transient fast channel and steady-state slow channel) and the reference signal in this invention. The error generated by the traditional hard-switching architecture (which directly uses the observer's estimated rotation angle plus the rotation angle calculated by the auxiliary equation, and outputs the average after summing, without filtering) compared to the reference signal The comparison chart.

[0086] like Figure 5 and Figure 6 As shown, under the dual impact of sharp steering and sensor failure, the EKF algorithm reveals a significant estimation blind zone, with internal error accumulation leading to estimation bias in the 4s-8s range. In contrast, this invention, thanks to its robustness to complex disturbances, maintains excellent convergence speed, accurately reconstructs complex disturbances, and does not induce sliding mode chattering.

[0087] Figure 7 This is a comparison chart of tracking error and output torque under fishing hook working conditions, where (a) is the fault-tolerant rotation angle of the present invention. EKF fault-tolerant rotation angle (a) Comparison chart; (b) Output torque of the present invention With EKF output torque The comparison chart.

[0088] from Figure 7 It can be seen that, directly affected by the divergence of the EKF estimate, Figure 7 In (b) of the paper, the motor torque exhibits abrupt changes and spikes, a phenomenon that can easily lead to steering instability in practical engineering. Thanks to the deep coupling of pre-positioned soft isolation and sliding mode fault-tolerant control law, this invention achieves a smooth transition of the reconstructed signal, making the output motor torque stable and continuous, thus avoiding the risk of instability under high dynamic alternation.

[0089] Traditional extended Kalman algorithms inevitably induce step transitions in the reconstructed signal; however, this invention, relying on a pre-amplified soft isolation channel based on Logistic attenuation characteristics, effectively reduces the error shear slope and achieves a smooth transition of the fault-tolerant feedback signal. Through the smooth fusion of multi-source data, this invention maintains excellent continuity and smoothness in the output motor torque curve, enabling a safe switch from normal mode to fault-tolerant mode.

[0090] In summary, the sensor-tolerant control method for drive-by-wire vehicles based on steering motor torque regulation according to the above embodiments has the following beneficial effects: (1) The present invention constructs a dynamic model of SBW system containing complex nonlinear disturbances. By equating the nonlinear friction of the actuator, gear meshing error, road load change, low-frequency time-varying drift, high-frequency harmonic fluctuation and random environmental noise as integrated disturbance terms and incorporating them into the system state space model, the model's representation accuracy of the real working conditions is effectively improved, providing a more accurate model basis for subsequent observer design and fault-tolerant control.

[0091] (2) This invention designs a superspiral sliding mode observer, which effectively solves the problem of state estimation under complex disturbances. The observer uses unmodeled states such as nonlinear friction as integrated disturbance terms for real-time estimation and compensation. By utilizing the finite-time convergence characteristics of the superspiral sliding mode error compensation algorithm, it achieves high-precision real-time estimation of system state variables and equivalent integrated disturbances, thereby improving the state estimation accuracy and robustness of the SBW system.

[0092] (3) This invention proposes a fault-tolerant control architecture based on multi-source data fusion. This architecture constructs a pre-soft isolation and exponential smooth transition mechanism. That is, it generates soft isolation weights through the Logistic function of the transient fast channel to achieve fast response, and achieves false alarm avoidance through the Schmitt hysteresis discriminator of the steady-state slow channel. In the signal reconstruction, a fusion ratio of 9:1 is adopted, which can achieve smooth reconstruction of the steering angle signal under complex disturbance and sensor failure conditions, reduce the step impact of steering motor torque caused by diagnostic delay, false alarm and hard switching, and improve the tracking stability, fault tolerance safety and smoothness of the steer-by-wire system.

[0093] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A sensor-tolerant control method for drive-by-wire vehicles based on steering motor torque regulation, characterized in that, include: Step S1: Based on the vehicle's two-degree-of-freedom dynamic equations, the steering motor's dynamic equations, the front wheel's dynamic equations, and the transmission relationship, establish a dynamic model of the steer-by-wire system; Step S2: Based on the dynamic model of the steer-by-wire system established in Step S1, the nonlinear friction of the actuator, gear meshing error, road load change, low-frequency time-varying drift, high-frequency harmonic fluctuation and random environmental noise are equivalent to integrated disturbances, and a state space model of the steer-by-wire system is established. Step S3: Based on the state space model of the steer-by-wire system established in step S2, construct a super-spiral sliding mode observer, define the state estimation error and the integrated disturbance estimation residual of the super-spiral sliding mode observer, and thus construct the error state equation. Based on the error state equation, use the super-spiral sliding mode error compensation algorithm to estimate the state variables and integrated disturbance of the steer-by-wire system in real time, and finally output the observer's estimated steering angle and disturbance estimate. Step S4: Obtain the actual rotation angle measured by the sensor, and combine it with the estimated rotation angle obtained in step S3, as well as the kinematic auxiliary rotation angle calculated based on the vehicle's yaw rate, to form multi-source rotation angle data; Step S5: Using the multi-source rotation angle data, calculate the observer transient residual between the sensor measured rotation angle and the observer estimated rotation angle, and the auxiliary transient residual between the sensor measured rotation angle and the kinematic auxiliary rotation angle, respectively. Extract the minimum transient residual using the minimum value operator, and then use the Logistic function to perform nonlinear mapping on the minimum transient residual to generate sensor soft isolation weights. Step S6: Filter the transient residual of the observer and the auxiliary transient residual using a first-order discrete low-pass filter to obtain the steady-state residual of the observer and the auxiliary steady-state residual. Then, extract the minimum steady-state residual using a minimum value operator. Finally, use a Schmitt hysteresis discriminator to determine the threshold of the minimum steady-state residual and update the sensor fault flag bit. Step S7: Based on the sensor fault flag bit updated in step S6, the observer estimated rotation angle and the kinematic auxiliary rotation angle, generate a fused rotation angle value according to a preset fusion ratio, and then combine the sensor soft isolation weight generated in step S5 with the sensor measured rotation angle to perform weighted reconstruction and generate a fault-tolerant rotation angle value. Step S8: Reconstruct the system state variables using the fault-tolerant angle value, construct a sliding mode fault-tolerant control law based on the reconstructed state variables, and finally calculate and output the steering motor control torque through the sliding mode fault-tolerant control law.

2. The sensor-tolerant control method for drive-by-wire vehicles based on steering motor torque regulation according to claim 1, characterized in that, In step S2, when the nonlinear friction of the actuator, gear meshing error, road load variation, low-frequency time-varying drift, high-frequency harmonic fluctuations, and random environmental noise are equated as integrated disturbances, the following equation is satisfied. in, Indicates integrated disturbance. Represents the Coulomb friction coefficient. Represents a symbolic function. The angular velocity of the front wheel's steering angle. It is a composite perturbation function. For low-frequency disturbance coefficients, For high-frequency disturbance coefficients, This represents the environmental noise disturbance factor. This is the low-frequency disturbance frequency. For high-frequency disturbance frequencies, For random environmental noise, For time.

3. The sensor-tolerant control method for drive-by-wire vehicles based on steering motor torque regulation according to claim 2, characterized in that, Step S2, the process of establishing the state-space model of the steer-by-wire system specifically includes: Considering external disturbances to the vehicle, and taking the front wheel steering angle, front wheel steering angular velocity, vehicle center of gravity sideslip angle, and yaw rate as system state variables, a state-space model of the steer-by-wire system reflecting the vehicle's physical dynamics is established, expressed as: in, Indicates the front wheel steering angle. Angular acceleration representing the steering angle of the front wheel. This represents the sum of the front wheel aerodynamic trail and the mechanical trail. For the front wheel lateral stiffness, For rear wheel lateral stiffness, For the equivalent front wheel rotational inertia, The equivalent friction coefficient, The sideslip angle is the angle of the center of mass. The angular velocity of the center of mass deflection. This is the distance from the front axle to the center of gravity. This is the distance from the rear axle to the center of gravity. For the longitudinal speed of the vehicle, The yaw rate is angular velocity. This is the yaw acceleration. The steering gear ratio, For the steering motor to output torque, For vehicle quality, This is the moment of inertia of the entire vehicle about the vertical direction.

4. The sensor-tolerant control method for drive-by-wire vehicles based on steering motor torque regulation according to claim 3, characterized in that, In step S3, the superhelical sliding mode observer satisfies the following equation: in, These are estimates of the state variables. for The first derivative, To provide an estimate of the output variable, For integrated disturbance The estimated value, For sliding mode variables, For the system control law, The observer feedback gain matrix, , , , For the system matrix; Among them, matrix Satisfying Hurwitz stability, Indicates the selection vector. , Table transpose; The expression for the system matrix is: 。 5. The sensor-tolerant control method for drive-by-wire vehicles based on steering motor torque regulation according to claim 4, characterized in that, In step S3, the state estimation error and the integrated perturbation estimation residual of the superspiral sliding mode observer are defined as follows: in, For state estimation error, To integrate the disturbance estimation residuals, For state variables, This is the error in the front wheel steering angle estimation. The angular velocity estimation error for the front wheel steering angle. This is the error in estimating the centroid sideslip angle. This represents the estimation error of the yaw rate.

6. The sensor-tolerant control method for drive-by-wire vehicles based on steering motor torque regulation according to claim 5, characterized in that, In step S3, combining the state-space model of the steer-by-wire system and the super-helical sliding mode observer, the error state equation is constructed, and its expression is: in, yes The first derivative; Then define the equivalent perturbation. Estimated value of equivalent disturbance Equivalent disturbance error ; Therefore, the superspiral sliding mode error compensation algorithm used in step S3 satisfies the following equation: in, , For observer gain parameters, , , For observer intermediate variables, It is the first derivative of the intermediate variable of the observer.

7. The sensor-tolerant control method for drive-by-wire vehicles based on steering motor torque regulation according to claim 6, characterized in that, In step S5, the observer transient residual and the auxiliary transient residual are calculated using the following formula: in, For the transient residual of the observer, To assist in transient residuals, The actual rotation angle measured by the sensor. Estimate the rotation angle for the observer. To assist in kinematic rotation; Then, the minimum transient residual is extracted using the minimum value operator, expressed as: in, It is the minimum transient residual; Set an absolute dead zone , To tolerate the warning threshold, and then extract the effective residual, the sensor soft isolation weight is generated, expressed as: in, For effective residuals, For sensor soft isolation weights, The shape factor of the Logistic function that determines the isolation rate. It is a natural exponential function.

8. The sensor-tolerant control method for drive-by-wire vehicles based on steering motor torque regulation according to claim 7, characterized in that, In step S6, a first-order discrete low-pass filter is used to filter the observer transient residual and the auxiliary transient residual respectively, as expressed by: in, These are the filter coefficients. The time constant of the low-pass filter. The sampling period is For discrete sampling time indexes, for The steady-state residual of the observer at time t. for The auxiliary steady-state residual at each moment, for The steady-state residual of the observer at time t. for The auxiliary steady-state residual at each moment, for The transient residual of the observer at time t, for Auxiliary transient residuals at each moment; Then, the minimum steady-state residual is calculated using the minimum value operator, expressed as: in, To minimize the steady-state residual, For the steady-state residuals of the observer, To assist in steady-state residuals; Then, the sensor fault flag is updated using a Schmitt hysteresis discriminator, expressed as: in, The threshold for diagnosis for The sensor fault flag bit at any given time. for Sensor fault flag bit at any time, sensor fault flag bit This indicates that the sensor has malfunctioned. This indicates that the sensor is working properly.

9. The sensor-tolerant control method for drive-by-wire vehicles based on steering motor torque regulation according to claim 8, characterized in that, In step S7, the fault-tolerant rotation angle value is calculated using the following formula: in, To integrate corner values, This is the fault-tolerant rotation angle value.

10. The sensor-tolerant control method for drive-by-wire vehicles based on steering motor torque regulation according to claim 9, characterized in that, Step S8 specifically includes: The fault-tolerant rotation angle value generated in step S7 Using the reliable front wheel steering angle information of the system at the current moment, and combined with the state variables estimated by the super-helical sliding mode observer, the reconstructed state variables are constructed. ; Then define the feedback tracking error. The expression is: in, This is the desired reference state; Redesign the sliding surface The expression is: in, Here is the sliding surface parameter matrix; Constructing an exponential reaching law using a saturation function The expression is: in, For the gain of the switching item, , For exponentially approaching gain, , It is a saturation function; Saturation function The definition of is: in, This is the input quantity for the saturation function; Finally, a sliding mode fault-tolerant control law is constructed. The expression is: in, This represents the expected rate of change of the reference state.

Images