Research On Fault Diagnosis And Fault-Tolerant Strategy For Redundant Steering System Of Autonomous Vehicles Considering Functional Safety

Driven by favorable policies related to autonomous driving and the research and development boom in the industry,richer and safer automated driving functions will become the core competitiveness of enterprises to seize the future automobile market.With the improvement of automated driving level,the driver’s participation in dynamic driving tasks gradually decreases,and more often it relies on the automated driving system to complete dynamic driving tasks independently.In this case,the functional integrity of the relevant system becomes particularly important.As an essential core component of vehicle,steering system has gradually changed from basic power assistance to automatic steering.In order to achieve the requirements of automated driving system and improve the reliability and safety of steering system,it is necessary to consider functional safety in its architecture design and control strategy development.In accordance with the functional safety analysis process of ISO26262 standard and focusing on the automatic steering function,this study carries out functional safety analysis for the steering system of autonomous vehicles and proposes corresponding fault diagnosis and fault-tolerance strategies,so as to realize the monitoring of the functional state and ensure that the vehicle can enter the safety state after the failure of the sensor,control and execution unit in the system.Overall,The main contributions of this study are as follows:(1)Functional Safety analysis of redundant steering system for autonomous vehiclesAccording to the ISO 26262 standard,the functional safety analysis of steering system for autonomous vehicles is carried out.Firstly,the requirements of automatic steering function and operation scenarios are clarified,relevant items are defined,and the initial architecture and signal interaction relationship are given.Then hazard analysis and risk assessment are carried out to derive the hazard events and give the Automotive Safety Integrity Level(ASIL)and safety goals.After that,the functional safety goals are decomposed into functional safety requirements and technical safety requirements to guide the design of software and hardware.Based on this,the hardware redundancy design and the update of control architecture are completed.(2)Sensor fault diagnosis and fault-tolerance strategy for redundant steering systemThe fault diagnosis and fault-tolerance strategies of the sensors are designed according to the technical safety requirements of the sensors.First of all,the representation of the sensor fault types at the signal level is analyzed and the fault models are established.Then,the support vector regression method is used to design the Angle estimator.Combined with hardware and analytic redundancy,use the majority voting method to realize the fault diagnosis of the angle sensor.When the faulty sensor is confirmed,it switched to the remaining normal sensors to achieve fault-tolerance.Considering the characteristics of the current sensor,the coordinate transformation and Nonlinear Auto-Regressive model with e Xogenous inputs neural network are combined to realize the sigle-step and multi-step prediction of current sensor.The fault diagnosis of the current sensor is realized by the residual sequence between the predicted current value and the measured value of sensor,and the judgment of the constant offset fault and constant gain fault is given.When the fault of the current sensor is not serious and easy to locate,fault-tolerance compensation strategy is implemented according to the extracted fault type and fault value.Finally,the effectiveness of the designed sensor fault diagnosis and faulttolerance switching strategy is verified on the co-simulation platform.(3)Fault diagnosis and fault-tolerance strategy of control and execution units for redundant steering systemFirstly,according to the actual structure of the redundant steering system,the mechanical model of the steering system and the Double-winding Permanent Magnet Synchronous Motor(DW-PMSM)model are built.Based on this,the automatic steering control algorithm is designed,and the corresponding controller fault diagnosis strategy is given.Then,the common winding faults of the DW-PMSM are analyzed,and the corresponding fault models are established.The third harmonic signal of the phase current is extracted by the Short Time Fourier Transform to realize the fault diagnosis of the motor winding.Finally,the fault-tolerant switching and degradation logic of the redundant steering system are designed.The effect of automatic steering control and the effectiveness of fault diagnosis and fault-tolerance strategy of motor winding are verified on the co-simulation platform.(4)Autonomous vehicles redundant steering system fault injection test verificationFirst of all,according to the hazard scenarios extracted from the functional safety analysis and combined with the steering system functional safety standards,the fault injection test scheme and fault injection test cases are given.Then,based on the CarSim/Simulink cosimulation platform,the fault injection simulations are carried out to verify the designed fault diagnosis and fault-tolerance strategy.The test results show that the designed safety mechanisms can effectively detect faults and correctly realize fault-tolerance according to the demand.In order to further verify the effectiveness of the designed strategy,the Hardware-inthe-loop test bench of the steering system is built,and the automatic steering function test and fault injection test are completed on the bench.The test results show that automatic steering angle tracking effect is good and the fault diagnosis and fault-tolerant strategy established in this paper can diagnose and tolerate faults quickly after fault injection,make the vehicle enter the corresponding safe state and avoid the occurrence of hazardous events,so as to improve the the reliability and safety of the steering system.