Steer-by-wire emergency response method

Abstract

The invention discloses a steer-by-wire emergency response method which comprises the following steps: in a vehicle driving process, detecting whether a main steering system fails or not in real time, and controlling the vehicle to steer through the main steering system when the main steering system does not fail; when the main steering system runs, the self-learning module generates dynamic compensation parameters according to the running parameters of the main steering system and the backup power steering system; the backup steering system controls the vehicle to steer based on the dynamic compensation parameters when the main steering system breaks down. The state of the main steering system is detected in real time, dynamic compensation parameters are generated through the self-learning module, non-delay switching control is achieved when the system breaks down, and meanwhile errors caused by mechanical losses and working condition changes are overcome through a dynamic parameter compensation mechanism; the method effectively solves the problems of control delay and insufficient static compensation precision in the prior art, and has the remarkable advantages of improving the response speed and operation safety of the system.

Description

Technical Field

[0001] The present invention belongs to the technical field of vehicle steering control, and in particular relates to a wire-controlled steering emergency response method. Background Art

[0002] With the rapid development of autonomous driving technology, steer-by-wire systems have become the core actuator of intelligent connected vehicles. However, existing technologies face significant technical bottlenecks when the primary steering system fails. First, there is a significant control delay during system switching. When the primary steering system fails, the redundant controller must take over as the backup steering system. However, hardware communication delays typically exceed 100 milliseconds. Combined with parameter cold start issues, this often leads to transient overshoot during vehicle steering, manifesting as steering wheel judder or understeer / oversteer, among other safety hazards. Second, the static compensation mechanism used in existing backup systems has inherent flaws. Relying on factory-calibrated fixed parameters or simple deviation compensation methods, it cannot effectively adapt to dynamic changes in actual operation, such as motor carbon brush wear and changes in mechanical clearance caused by temperature fluctuations. Consequently, compensation accuracy degrades over time. These issues severely limit the reliability and safety of steer-by-wire systems, particularly in critical scenarios such as high-speed driving, where they can pose serious safety risks.

[0003] In view of the above problems, the existing technology is in urgent need of improvement. Summary of the Invention

[0004] The purpose of the present invention is to address the deficiencies of the above-mentioned background technology and to provide a wire-controlled steering emergency response method, which has the advantages of effectively reducing system switching delay, improving compensation accuracy and operational reliability.

[0005] The technical solution adopted by the present invention is: a wire-controlled steering emergency response method, which detects in real time whether the main steering system is faulty during vehicle driving, and controls the vehicle steering through the main steering system when the main steering system is not faulty; when the main steering system is running, the self-learning module generates dynamic compensation parameters according to the operating parameters of the main steering system and the backup dynamic steering system; when the main steering system fails, the backup steering system controls the vehicle steering based on the dynamic compensation parameters.

[0006] Furthermore, the present invention also proposes that when the main steering system is running, the self-learning module controls the backup steering system to simulate the steering action of the main steering system, and generates dynamic compensation parameters by comparing the actual control parameters of the main steering system and the backup steering system.

[0007] Furthermore, the present invention also proposes that the steering action is a reciprocating rotation at a set angle.

[0008] Furthermore, the present invention also proposes that the dynamic compensation parameters include steering angle deviation and motor torque correction value.

[0009] Furthermore, the present invention also proposes that when the main steering system fails, the backup steering system controls the vehicle steering based on the dynamic compensation parameters, including: superimposing the dynamic compensation parameters into the control parameters of the backup steering system to control the vehicle steering.

[0010] Furthermore, the present invention also proposes that when the dynamic compensation parameter exceeds a threshold value, the backup steering system limits the vehicle parameters when controlling the vehicle steering.

[0011] Furthermore, the present invention also proposes that limiting vehicle parameters includes: limiting the vehicle speed to be lower than a speed setting value and / or limiting the steering angle to be lower than an angle setting value.

[0012] Furthermore, the present invention also proposes that the self-learning module is any one of an LSTM network model, an RNN network model, an SVR model, and a BP neural network model.

[0013] Furthermore, the present invention also proposes that when the main steering system fails and the backup steering system controls the vehicle steering, a warning signal is sent to the driver through the instrument panel and / or audio equipment to prompt the steering system to enter emergency mode.

[0014] Furthermore, the present invention also proposes that when the main steering system fails and the backup steering system controls the steering of the vehicle, the vehicle speed and steering angle are limited.

[0015] The beneficial effects of the present invention are:

[0016] The present invention provides a wire-controlled steering emergency response method, which realizes delay-free switching control in the event of a system failure by detecting the status of the main steering system in real time and using a self-learning module to generate dynamic compensation parameters. At the same time, the dynamic parameter compensation mechanism is used to overcome the errors caused by mechanical losses and changes in working conditions, effectively solving the problems of control delay and insufficient static compensation accuracy in the prior art, and has the significant advantage of improving system response speed and operational safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 Flowchart of the present invention. DETAILED DESCRIPTION

[0018] The following is a further description of specific embodiments of the present invention in conjunction with the accompanying drawings. It should be noted that the description of these embodiments is intended to facilitate understanding of the present invention and does not constitute a limitation of the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

[0019] In existing technologies, steer-by-wire systems, as core components of intelligent connected vehicles, have a significant impact on driving safety. Traditional systems rely on redundant controllers for backup when the primary steering system fails. However, hardware communication delays can lead to overshoot in the steering transient response, manifesting as steering wheel jitter or steering angle deviation. Furthermore, the backup system utilizes factory-calibrated fixed compensation parameters, which are unable to cope with dynamic scenarios such as motor carbon brush wear and changes in mechanical clearance due to temperature fluctuations. Consequently, compensation accuracy significantly decreases over time.

[0020] To address these issues, the first step is to address the dynamic instability that occurs during active / standby system switching. Traditional parameter compensation methods lack real-time adjustment capabilities, making it difficult to quickly adapt to actual operating conditions during sudden failures. Analysis has shown that continuously training the compensation model while the primary system operates normally can effectively eliminate cold start delays. Secondly, to address hardware performance degradation, a method is proposed to acquire real-time data through the synchronous operation of the primary and standby systems and establish a dynamic parameter mapping relationship. This adaptive compensation mechanism addresses the challenges of data collection and model updates during the coordinated control of the primary and standby systems.

[0021] To achieve the above objectives, the present invention provides a steer-by-wire emergency response method, such as Figure 1 As shown, during the vehicle's driving process, the main steering system is detected in real time to see if it is faulty. When there is no fault, the automatic driving controller controls the vehicle steering through the main steering system; when the active system is running, the self-learning module generates dynamic compensation parameters based on the operating parameters of the main and backup systems; when the active system fails, the backup system controls the vehicle steering based on the dynamic compensation parameters.

[0022] Among them, real-time detection refers to the continuous monitoring of the steering motor's current fluctuations, angle sensor signals, vehicle speed and other state parameters through the vehicle's controller local area network. This can be achieved by using a fault diagnosis algorithm, such as a residual analysis method based on Kalman filtering. The self-learning module generates dynamic compensation parameters by using a machine learning model to learn the control differences between the primary and backup systems online. Specifically, a recursive neural network can be used to process time series data, and the compensation coefficient is generated by comparing the deviation between the active steering angle and the backup system's simulated angle. The backup system controls vehicle steering by injecting the learned parameters into the steering control command in real time. This can be achieved by superimposing the compensation value on a PID controller.

[0023] It's understandable that during normal vehicle operation, while the primary steering system performs steering maneuvers, the backup steering system, under the control of the self-learning module, simultaneously simulates the same steering motion. Parameters such as torque output and steering angle from both systems are collected in real time, and a dynamic compensation model is established through time series analysis. In the event of a sudden failure in the primary steering system, the backup steering system directly calls upon the newly generated compensation parameters to correct the steering motor's output characteristics, without waiting for parameter initialization. This mechanism effectively avoids the control vacuum period associated with traditional system switching and adapts to nonlinear changes caused by mechanical component wear through continuous parameter updates.

[0024] Compared to existing technologies, traditional backup systems use fixed compensation tables, making them unable to automatically correct steering angle deviations when mechanical clearance increases. This solution, through an online learning mechanism, dynamically adjusts the compensation coefficient based on the actual motor output characteristics. While existing technologies require manual calibration of compensation parameters, this solution achieves self-updating parameters through collaborative training of the primary and backup systems. Traditional methods require reinitialization of control parameters during system switching, while this solution achieves seamless transition through pre-trained models.

[0025] Through the above-mentioned technical solution, the present invention achieves smooth switching control between the primary and backup steering systems, eliminating the steering jitter caused by communication delays in traditional systems. The continuous update mechanism of dynamic compensation parameters enables the backup system to adaptively adjust steering torque output, compensating for transmission clearance errors caused by mechanical component wear. In scenarios such as low temperatures causing changes in steering mechanism rigidity, the self-learning model can modify the torque output curve in real time to maintain steering control accuracy.

[0026] In some embodiments, the present invention further proposes that when the main steering system is running, the self-learning module controls the backup steering system to simulate the steering action of the main steering system, and generates dynamic compensation parameters by comparing the actual control parameters of the main steering system and the backup steering system.

[0027] Among them, the backup steering system simulates the steering action of the main steering system, which means that the backup steering system performs the same mode of steering operation according to the control signal output by the main system. Specifically, it can be achieved by using signal replication or motion trajectory tracking algorithm, so that the backup system can continue to perform dynamic training in a non-fault state.

[0028] The actual control parameters refer to the steering motor torque, steering angle feedback value or current response curve, which can be collected by sensors and transmitted to the controller to quantify the control differences between the two systems under the same action.

[0029] Among them, dynamic compensation parameter generation refers to calculating the control deviation between the active and standby systems through an error function. Specifically, it can be implemented using an online recursive least squares method or a gradient descent algorithm to construct a real-time updated compensation mapping table to eliminate hardware characteristic differences.

[0030] It is understood that during the normal operation of the main steering system, the backup steering system is forcibly triggered to perform steering actions completely synchronized with the main system. For example, when the main system performs a steering operation at a specific angle, the backup system receives control instructions of the same frequency and amplitude and simulates the action. The actual torque output values and angle feedback signals of the main and backup systems are collected in real time and input into the self-learning module. By calculating the absolute error and rate of change between the two, a compensation coefficient matrix related to the steering angle and vehicle speed is established. This matrix is stored in the controller's non-volatile memory. In the event of a failure in the main system, the control signal of the backup system will be superimposed with the correction value corresponding to the compensation coefficient.

[0031] Compared with existing technologies, traditional solutions rely on factory-set fixed compensation parameters and are unable to address torque attenuation caused by motor carbon brush wear or mechanical backlash offset caused by temperature changes. This solution, by continuously training the backup system during normal system operation, enables the compensation parameters to track mechanical transmission errors and electrical characteristic drift in real time, avoiding parameter cold start issues during hardware switching. For example, in scenarios with motor response delays, dynamic compensation parameters can automatically adjust the PID control parameters of the current loop, ensuring that the backup system's steering action remains synchronized with the primary system.

[0032] Through the above-mentioned technical solution, the present invention eliminates control signal jumps when the backup system takes over, reducing transient steering overshoot. For example, if the primary system suddenly fails, the backup system can directly call upon trained compensation parameters to achieve a smooth transition, avoiding steering wheel judder or sudden changes in steering angle. Furthermore, the dynamic compensation mechanism adaptively compensates for changes in gear backlash caused by ambient temperature fluctuations, ensuring that the linearity of steering torque output is consistent with that of the primary system.

[0033] The present invention further proposes that the steering action is a reciprocating rotation at a set angle.

[0034] The set angle refers to a steering range pre-set to multiple typical angles. This can be achieved by combining a small 5-degree correction steering angle with a larger 30-degree lane change steering angle to cover typical steering conditions during vehicle operation. Reciprocating rotation, which periodically alternates between clockwise and counterclockwise steering, can be achieved by generating a sine wave or triangular wave steering angle command signal from the motor controller. This signal is used to capture the dynamic response differences of the steering system during forward and reverse motion.

[0035] Specifically, when performing simulated actions, the backup steering system is constrained to perform periodic reciprocating motions according to a preset angle sequence. For example, the angle sequence may include three gradient values of 5 degrees, 15 degrees, and 30 degrees, and each gradient value corresponds to the typical requirements of different steering scenarios. At each angle gradient, the actuator rotates alternately clockwise and counterclockwise at a fixed frequency, and collects nonlinear changes in parameters such as motor torque and gear clearance through a complete motion cycle. This constrained action mode avoids the blind spots in parameter collection caused by uneven angle distribution in random steering tests, allowing the dynamic compensation parameters to cover the full range of operating conditions from fine-tuning to large-scale steering.

[0036] Compared to existing technologies, traditional backup systems collect compensation parameters only through a single random steering motion, which can easily lead to compensation model failure due to the test angle not covering the actual operating conditions. However, this invention establishes a systematic parameter learning framework by standardizing angle sequences and periodic motion patterns. This framework can fully capture the hysteresis effect and backlash error of the steering mechanism in both forward and reverse motion, resolving the technical drawback of incomplete data samples in traditional cold start compensation.

[0037] Through this technical solution, the present invention effectively eliminates dynamic compensation parameter deviations caused by random steering movements, enabling the backup system to accurately replicate the control characteristics of the primary steering system. By constraining the angular range and motion period of reciprocating rotation, the compensation parameters fully cover the vehicle's actual steering requirements, preventing understeer or overcompensation.

[0038] In some embodiments, the present invention further proposes that the dynamic compensation parameters include steering angle deviation and motor torque correction value.

[0039] Steering angle deviation refers to the difference between the actual output angles of the primary and backup steering systems. This is achieved by using a steering sensor to collect the steering angle signals from both systems in real time, filtering the signals to eliminate noise interference, and then calculating the difference. This parameter is used to eliminate angular deviations between the two systems caused by mechanical backlash or transmission errors.

[0040] The motor torque correction value is a torque compensation value dynamically adjusted based on load changes or motor performance degradation. This is achieved by using a current sensor to monitor the actual output torque of the backup steering motor. This is combined with a preset torque-current relationship model to generate a real-time correction value via a PID controller. This parameter is used to compensate for torque output deviations caused by decreased motor efficiency or external load fluctuations.

[0041] Specifically, steering angle deviation is calculated by comparing the real-time steering angle data of the active steering system with that of the backup steering system. This angle compensation is then input into the backup steering controller, ensuring that both systems maintain consistent steering trajectories under the same input command. The motor torque correction dynamically adjusts the drive current of the backup steering motor based on its actual load state, ensuring that the output torque always matches the calibrated torque of the primary steering system. By simultaneously correcting the steering angle and drive torque, the backup steering system can adjust compensation parameters based on real-time operating conditions, preventing a decrease in control accuracy due to hardware aging or environmental changes.

[0042] Compared to existing technologies, traditional backup systems rely solely on fixed compensation parameters preset at the factory, making them unable to address efficiency degradation caused by motor carbon brush wear or changes in mechanical clearance due to temperature fluctuations. This solution, however, calculates steering angle deviation and motor torque correction values in real time, forming a dynamic closed-loop compensation mechanism. This ensures that the backup system maintains control consistency with the primary steering system despite hardware performance changes or environmental interference.

[0043] Through the above technical solution, the present invention solves the problem of steering control inaccuracy caused by the inability of static compensation parameters of traditional backup systems to adapt to dynamic scenarios. Through two-dimensional dynamic compensation of steering angle and driving torque, the control accuracy and stability of the backup system under different working conditions are effectively improved.

[0044] In some embodiments, the present invention further proposes that when the primary steering system fails, the backup steering system controls vehicle steering based on dynamic compensation parameters, including superimposing the dynamic compensation parameters into the control parameters of the backup steering system to control vehicle steering.

[0045] The dynamic compensation parameter refers to the steering control correction generated in real time by the self-learning module. This can be achieved using steering angle deviation and motor torque correction values. The steering angle deviation is used to compensate for the position error between the backup system and the main system's steering actuator, while the motor torque correction value is used to match the torque output characteristics of the main and backup systems. During normal operation of the main steering system, the self-learning module continuously collects the steering angle, motor current, and vehicle speed data of the main steering system, and generates dynamic compensation parameters through time series modeling. These parameters can reflect the current real-time status of the vehicle, such as changes in mechanical clearance and motor performance degradation.

[0046] Adding the dynamic compensation parameters to the backup steering system's control parameters refers to integrating the dynamic compensation parameters with the backup system's control signals. This can be achieved by linearly superimposing the control signals or injecting the control algorithm into the system through a feedforward channel. When the backup system takes over steering control, the dynamic compensation parameters are directly added to the steering angle command signal or motor torque command signal. This ensures that the backup system's initial control variables already include the correction parameters for the current operating conditions, eliminating the need for parameter initialization.

[0047] Specifically, when the primary steering system is operating normally, the self-learning module synchronously collects the primary steering system's steering angle, motor torque, and vehicle speed data, while simultaneously controlling the backup steering system to perform the same steering maneuvers as the primary system. By comparing the actual steering angles and motor currents of the primary and backup systems, it generates a steering angle deviation compensation coefficient and a torque correction gain. If a failure in the primary steering system is detected, the backup system immediately takes over control. At this point, the steering control module adds the steering angle deviation compensation coefficient to the target steering angle calculation loop and simultaneously injects the torque correction gain into the motor driver's current closed-loop control. For example, if the steering angle deviation is +1.5 degrees, the backup system's target steering angle command will be corrected to the original command value plus 1.5 degrees. If the motor torque correction value is 0.9 times the current value, the drive current command will be scaled accordingly. This allows the backup system to directly inherit the dynamic adaptation parameters of the primary system before the failure, eliminating control command jumps caused by hardware communication delays.

[0048] Compared with existing technologies, traditional redundant steering systems require reinitialization of the backup system's control parameters when switching, such as calling preset fixed compensation values from memory or estimating parameters based on short-term data before the fault, resulting in initialization time that can reach over 100ms. However, this solution uses real-time superposition of dynamic compensation parameters, ensuring that the backup system's control parameters already include the optimal correction value for the current vehicle state at the moment of fault switching, thus avoiding the parameter calibration process during the cold start phase. The static compensation parameters in existing technologies cannot adapt to changes in torque characteristics caused by motor carbon brush wear, while the dynamic compensation parameters continuously track such hardware state changes through online learning. For example, when motor efficiency decreases, the torque correction value will adaptively increase to maintain the same steering torque output.

[0049] Through the above technical solution, the present invention solves the control delay problem caused by the cold start of the backup system parameters when the main steering system fails. The control command switching time is shortened to less than 20ms, and the steering angle fluctuation amplitude is reduced by more than 60%. At the same time, the dynamic compensation parameters correct the torque output curve of the backup system in real time, so that the steering wheel angle error is controlled within the range of ±0.5 degrees, avoiding vehicle instability caused by oversteering or understeering. In the hardware aging scenario, the self-learning module continuously updates the compensation parameters. For example, when the steering mechanism clearance increases by 0.3mm due to wear, the steering angle deviation parameter automatically increases the corresponding compensation amount to ensure that the steering control accuracy does not decay with time.

[0050] In some embodiments, the present invention further proposes that when the dynamic compensation parameter exceeds a threshold, the backup steering system limits the vehicle parameters when controlling the vehicle steering.

[0051] The dynamic compensation parameter exceeding the threshold refers to the dynamic compensation parameter exceeding the system's preset safety range. This can be achieved by using real-time sensor monitoring or by comparing the parameters output by the self-learning module with the preset threshold to determine whether the current compensation value is abnormal. Limiting vehicle parameters refers to imposing constraints on key parameters of the vehicle's operating state. This can be achieved by setting a speed limiter and a steering angle limiter in the backup steering system's control module to reduce steering control uncertainty when the compensation parameters are abnormal.

[0052] Specifically, after the backup steering system takes over vehicle steering control, the dynamic compensation parameters are input into the threshold judgment module in real time. When the dynamic compensation parameters output by the self-learning module exceed the preset threshold, it indicates that the current compensation value may exceed the safe range due to hardware aging, environmental interference, or system abnormality. At this time, the backup steering system automatically activates the vehicle parameter limitation mechanism, reducing the current vehicle speed to a pre-set safe speed range through the built-in speed limiter, and at the same time constraining the steering wheel rotation range through the steering angle limiter. This dual limitation mechanism can avoid sudden changes in steering torque caused by abnormal compensation parameters during the steering control process, thereby preventing vehicle trajectory deviation caused by overshoot of the steering transient response.

[0053] Compared to existing technologies, traditional backup systems only issue simple fault alarms when detecting compensation parameter anomalies, but are unable to proactively adjust vehicle operating conditions. The fixed threshold judgment method used in existing technologies struggles to adapt to dynamically changing vehicle operating conditions, resulting in steering control delays exceeding 100ms and prone to steering wheel jitter. This solution, through the synergy of real-time threshold judgment and parameter limiting, can complete anomaly detection and control strategy switching within 20ms. Furthermore, a graded steering angle limiting mechanism reduces steering wheel angle fluctuations by approximately 70%.

[0054] Through the above-mentioned technical solution, the present invention effectively suppresses steering system torque fluctuations and angle deviations when dynamic compensation parameters exceed a safe range, preventing vehicle steering judder or loss of control due to compensation failure. By limiting vehicle speed and steering angle in conjunction, the vehicle maintains a stable driving trajectory even under abnormal steering control conditions. This also avoids the steering delays that occur during traditional redundant system switching, reducing steering response time to one-third of the original system.

[0055] In some embodiments, the present invention further proposes that when the dynamic compensation parameter exceeds a threshold, the backup steering system limits vehicle parameters when controlling vehicle steering, limiting the vehicle speed to below a speed setting value and limiting the steering angle to below an angle setting value.

[0056] The speed setting value is the maximum speed allowed when the dynamic compensation parameters exceed the limits. This is achieved by using a vehicle speed sensor to collect wheel speed signals in real time and calculating a dynamic threshold using a preset algorithm in the electronic control unit. This parameter is used to reduce the impact of vehicle inertia on the steering system's response speed. The angle setting value is the maximum mechanical angle allowed for the steering actuator to rotate when the dynamic compensation parameters exceed the limits. This is achieved by using a steering angle sensor to monitor the steering column angle in real time and setting a safe angle range within the controller. This parameter is used to prevent the steering mechanism from mechanically exceeding the limits due to accumulated compensation errors.

[0057] Specifically, when the dynamic compensation parameter exceeds a threshold, the electronic control unit performs a closed-loop comparison between the real-time vehicle speed and the set speed value. When the vehicle speed exceeds the set speed, the drive motor output torque is limited to a correction value inversely proportional to the current vehicle speed, while linear deceleration control is applied via the braking system. The steering angle limit module is also activated simultaneously, dynamically adjusting the angle setpoint based on the current vehicle posture parameters. For example, on slippery roads, the setpoint may be reduced to 80% of that on dry roads. The two limiting parameters are synchronized via bus communication. Once the vehicle speed drops to a safe range, the steering angle setpoint gradually returns to its normal operating value according to a preset curve.

[0058] Compared to existing technologies, traditional backup systems only employ fixed speed reduction strategies or single angle limiting mechanisms when parameters exceed limits, failing to dynamically adjust the limit threshold based on real-time vehicle conditions. This solution establishes a dynamic correlation model between vehicle speed and steering angle, constructing a two-dimensional parameter limit matrix within the electronic control unit. This allows the vehicle to automatically match the optimal limit combination based on real-time driving conditions when compensating for parameter anomalies, avoiding secondary control issues caused by single parameter limitations.

[0059] Through the above technical solution, the present invention effectively solves the problem of vehicle lateral instability caused by steering system response delay and accumulated angle deviation when the dynamic compensation parameter exceeds the threshold value. On the premise of maintaining the mechanical safety of the steering mechanism, the risk of skidding under high-speed steering conditions is reduced through a dual parameter limitation mechanism, while avoiding mechanical interference failure of the steering actuator caused by exceeding the angle limit.

[0060] In some embodiments, the present invention further proposes that the self-learning module is a self-learning model, which can be any one of an LSTM network model, an RNN network model, an SVR model, and a BP neural network model; the learning mechanism of the self-learning model adopts an end-to-end sequential learning method.

[0061] Among them, the LSTM network model refers to the long-short-term memory network model, which can specifically model the long-term dependence of the dynamic parameters of the steering system through the memory units of time series data, and is used to capture the cumulative errors caused by motor wear. The RNN network model refers to the recurrent neural network model, which can specifically process the time-series change data of the steering system through a cyclically connected hidden layer, and is used to track the changes in mechanical clearance caused by temperature gradients in real time. The SVR model refers to the support vector regression model, which can specifically process the nonlinear relationship between the main steering system and the backup system through kernel function mapping, and is used to fit parameter drift under complex working conditions. The BP neural network model refers to the error back propagation neural network model, which can specifically optimize the generation accuracy of dynamic compensation parameters through a multi-layer perceptron structure, and is used to reduce the error accumulation of traditional step-by-step modeling. The end-to-end time series learning method refers to the overall mapping training directly from the input signal to the target compensation parameter. Specifically, the sequence-to-sequence model architecture can be used to avoid the information loss in traditional step-by-step processing, and is used to eliminate the parameter lag caused by hardware communication delay.

[0062] Specifically, when the main steering system is operating normally, real-time time series data of steering angle and motor torque are collected as input features and trained online using a selected machine learning model. For example, when using an LSTM network, the input layer receives steering action data within a continuous time window, the hidden layer dynamically adjusts the memory cell state through forget gates and input gates, and the output layer generates the corresponding steering angle deviation compensation. During the end-to-end training process, the loss function is defined as the mean squared error between the backup system's simulated steering action and the active system's actual response. The backpropagation algorithm automatically adjusts the network weights to minimize this error. The dynamic compensation model thus established can continuously adapt to torque attenuation caused by motor carbon brush wear, while automatically correcting mechanical transmission clearance errors caused by temperature changes.

[0063] Compared with existing technologies, traditional backup systems rely on fixed compensation parameters calibrated at the factory. They are unable to handle the nonlinear torque attenuation caused by motor carbon brush wear, and they also produce steering offset when temperature changes cause the mechanical clearance to increase. This solution uses a machine learning model to learn the dynamic characteristics of the steering system online. For example, when using the SVR model, the kernel function can map the nonlinear relationship between temperature sensor data and steering angle deviation to a high-dimensional space, thereby generating accurate real-time compensation parameters. For sudden hardware failure scenarios, the BP neural network quickly adjusts the hidden layer weights through error backpropagation, avoiding the lag of manual parameter adjustment compared to traditional PID compensation methods.

[0064] Through this technical solution, the present invention continuously optimizes compensation parameters under conditions of hardware wear and dynamic environmental changes. For example, when the contact resistance of the motor's carbon brushes increases due to long-term use, the LSTM network predicts the current attenuation based on historical torque data and generates a corresponding correction value. Similarly, when low temperatures in winter cause the steering mechanism to contract, the RNN network adjusts the steering angle compensation in real time based on temperature time series data. This eliminates the deviation between traditional static compensation parameters and the actual system state, ensuring that the backup steering system maintains the same steering response characteristics as the active system when taking over control.

[0065] In some embodiments, the present invention further proposes that when the main steering system fails and the backup steering system controls the vehicle steering, a warning signal is sent to the driver through the instrument panel and audio equipment to prompt the steering system to enter emergency mode.

[0066] The instrument panel refers to the display device inside the vehicle that displays vehicle status information. Specifically, it can be implemented as an LCD screen or mechanical pointer instrument, and it conveys warning information through flashing preset icons or text prompts. The audio device refers to the onboard audio output device. Specifically, it can be implemented as a speaker or buzzer, and it uses voice broadcasts or alarm sounds to indicate system status changes.

[0067] Specifically, upon detecting a failure of the primary steering system and the activation of the backup system, the system control module triggers a signal output command. The instrument panel simultaneously displays a warning steering fault icon, such as a red exclamation point and the text "Steering System Emergency Mode Activated." The audio system plays a warning tone three times, accompanied by a voice announcement: "Attention, the steering system has switched to backup mode." This solution leverages the output capabilities of existing onboard equipment, eliminating the need for additional hardware modules. Through dual-channel information transmission, the driver receives both visual and auditory warnings.

[0068] In some embodiments, the instrument panel warning signal can be configured to flash continuously until the driver manually acknowledges the warning. The audio system can be configured to emit different tones to differentiate warning levels. In cold environments, the interval between voice announcements can be extended to prevent audio distortion caused by icing on the speakers.

[0069] Compared to existing technologies, traditional solutions rely solely on a single instrument indicator light or lack a proactive warning mechanism, making it difficult for drivers to quickly identify system status changes during emergency steering switches. This solution, through multimodal interaction design, addresses the issue of delayed driver response caused by unclear or delayed warning signals, without incurring additional hardware costs.

[0070] Through the above technical solution, the present invention realizes the synchronous triggering of dual warning signals when the main steering system fails, so that the driver can immediately perceive the switch of vehicle control and adjust the steering operation force and driving speed in time to avoid the risk of misoperation caused by untimely information transmission.

[0071] In some embodiments, the present invention further proposes limiting vehicle speed and steering angle when the backup steering system controls the vehicle's steering when the primary steering system fails. For example, when a sharp bend, ramp, or obstacle is detected, the speed and steering angle limiting strategy is adjusted accordingly.

[0072] Among them, speed limit refers to reducing the control deviation of the backup system under high inertia conditions by setting a maximum speed threshold and dynamically adjusting the threshold. Specifically, this can be achieved by using a preset reference speed value and combining it with real-time road condition data to calculate the dynamic speed limit value. For example, the default maximum speed is set to 40km / h, and the speed limit is further reduced when a sharp turn is detected. This limit reduces the inertia of the vehicle's movement to avoid the risk of skidding due to the lag of compensation parameters. If an obstacle is detected ahead, the autonomous driving controller will appropriately reduce the speed and perform small steering operations to ensure that the vehicle can avoid it safely. Under complex road conditions, the controller will additionally call the lateral stability control function to adjust the differential speed of the front and rear wheels and the vehicle posture to ensure that the vehicle does not roll over or lose control when turning or avoiding.

[0073] Steering angle limitation dynamically constrains the mechanical range of the steering mechanism based on vehicle speed. This is achieved using a mapping table between vehicle speed and steering angle. For example, when the vehicle speed exceeds 30 km / h, the maximum steering angle is limited to 20 degrees, while when the vehicle speed is below 15 km / h, the steering angle is allowed to reach 45 degrees. This limitation controls the steering amplitude to prevent steering overshoot caused by delayed response of the backup system.

[0074] Specifically, when the primary steering system fails and the backup steering system is activated, the controller first reads the preset baseline speed limit and obtains real-time vehicle speed and steering angle data through on-board sensors. If the current vehicle speed exceeds the baseline speed limit, the braking system is immediately activated to coordinate deceleration until the speed drops to a safe range. At the same time, the maximum allowable steering angle is adjusted in real time based on vehicle speed: when the speed exceeds 30km / h, the steering angle is limited to within 20 degrees; when the vehicle enters a low-speed area, the steering angle is allowed to gradually increase to 45 degrees, and the actual steering angle is fed back in real time through the steering motor encoder. The controller performs closed-loop corrections to ensure that the steering action always remains within the preset angle range.

[0075] Compared with existing technologies, traditional methods rely solely on fixed speed limits or single steering angle constraints, failing to address the problem of sudden changes in vehicle yaw angle caused by insufficient dynamic compensation during high-speed steering. This solution utilizes a linked speed and steering angle limit mechanism to simultaneously reduce speed and steering angle at high speeds, effectively suppressing sudden changes in lateral acceleration while retaining wide-angle steering capability at low speeds, meeting emergency obstacle avoidance requirements.

[0076] Through this technical solution, the present invention can effectively suppress the accumulation of control errors in the backup steering system under complex operating conditions, preventing lateral vehicle instability caused by dynamic compensation lag. Speed limitation reduces steering system load fluctuations, while steering angle constraint prevents tire lateral forces from exceeding the grip limit. These two factors work together to ensure that the vehicle maintains a controllable lateral dynamic state during emergency steering.

[0077] In some embodiments, the present invention further proposes a post-fault recovery mechanism, including monitoring the recovery of the main steering system and returning control. After a main steering system failure, its control signals are blocked and unable to control the vehicle, but corresponding power supply and signal transmission information are available. This information can be used to monitor whether the main steering system has recovered.

[0078] Main steering system recovery monitoring means that after the backup steering system takes over, the system's fault monitoring module continues to monitor the main steering system's operating status. If the fault is resolved, the system will assess recovery conditions. Control return means that once the system determines that the main steering system has returned to normal, the backup system gradually returns steering control to the primary system. This process slowly and smoothly switches control back to the primary system to ensure a smooth transition during the steering process.

[0079] To implement the above-mentioned emergency response method, the present invention further provides a steer-by-wire emergency response system architecture, comprising:

[0080] The main steering system includes a fault monitoring module, a motor drive unit, a steering sensor, a control unit and other equipment.

[0081] The backup steering system is independent of the primary steering system and includes an independent motor drive unit, steering sensor, and control unit. The system is powered by a backup power supply and features an independent fault detection and warning module to ensure the independence and reliability of the backup system.

[0082] The self-learning module establishes a dynamic calibration model and generates dynamic compensation parameters according to the operating parameters of the main steering system and the backup dynamic steering system when the main steering system is running.

[0083] The automatic driving controller is connected to the main steering system and the backup steering system through a dedicated communication channel. When the main steering system is normal, it controls the vehicle steering through the main steering system. When the main steering system fails, it can immediately take over the control of the backup steering system to achieve vehicle direction control in the event of a fault.

[0084] It should be noted that the fault monitoring module monitors the signal transmission, power status, and actuator status of the main steering system to determine if there is a fault. Fault detection criteria include a series of thresholds, such as signal transmission interruption, insufficient power supply, and steering actuator abnormality. Once exceeded, a fault is considered. An early warning mechanism alerts the driver of potential steering anomalies via the instrument panel, audio, and other devices upon early detection of a system fault.

[0085] It should be noted that after a fault is detected, it is confirmed through repeated verification (for example, detecting multiple erroneous signals or continuous signal loss). If the fault is confirmed to be a steering failure, the backup steering system is triggered. After the system confirms the fault, it immediately cuts off the control path of the main steering system and sends a backup start signal to the autonomous driving controller. After the backup start signal is sent, the autonomous driving controller uses the backup steering device as the sole steering control channel and takes over the vehicle's steering control.

[0086] The above are only specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be covered by the scope of protection of the present invention. Matters not described in detail in this specification belong to the prior art known to those skilled in the art.

Claims

1. A steer-by-wire emergency response method, characterized by: During vehicle driving, the system detects whether the main steering system is faulty in real time and controls vehicle steering through the main steering system if the main steering system is not faulty. When the main steering system is running, the self-learning module generates dynamic compensation parameters based on the operating parameters of the main steering system and the backup dynamic steering system; When the primary steering system fails, the backup steering system controls the vehicle steering based on dynamic compensation parameters.

2. The steer-by-wire emergency response method according to claim 1, characterized in that: When the main steering system is running, the self-learning module generates dynamic compensation parameters according to the operating parameters of the main steering system and the backup dynamic steering system, including: When the main steering system is running, the self-learning module controls the backup steering system to simulate the steering action of the main steering system, and generates dynamic compensation parameters by comparing the actual control parameters of the main steering system and the backup steering system.

3. The steer-by-wire emergency response method according to claim 2, characterized in that: The steering action is a reciprocating rotation at a set angle.

4. The steer-by-wire emergency response method according to claim 1, characterized in that: The dynamic compensation parameters include steering angle deviation and motor torque correction value.

5. The steer-by-wire emergency response method according to claim 1, characterized in that: The backup steering system controls vehicle steering based on dynamic compensation parameters when the main steering system fails, including: The dynamic compensation parameters are added to the control parameters of the backup steering system to control the vehicle steering.

6. The steer-by-wire emergency response method according to claim 1, characterized in that: When the dynamic compensation parameter exceeds a threshold, the backup steering system limits the vehicle parameters when controlling the vehicle steering.

7. The steer-by-wire emergency response method according to claim 6, characterized in that: The limiting of vehicle parameters includes: limiting the vehicle speed to be lower than a speed setting value and / or limiting the steering angle to be lower than an angle setting value.

8. The steer-by-wire emergency response method according to claim 1, characterized in that: The self-learning module is any one of an LSTM network model, an RNN network model, an SVR model, and a BP neural network model.

9. The steer-by-wire emergency response method according to claim 1, characterized in that: When the primary steering system fails and the backup steering system controls the vehicle's steering, a warning signal is sent to the driver through the instrument panel and / or audio equipment, prompting the steering system to enter emergency mode.

10. The steer-by-wire emergency response method according to claim 1, characterized in that: When the main steering system fails and the backup steering system controls the vehicle steering, the vehicle speed and steering angle are limited.

Images