【作者】 朱敏；

【导师】 郁明； 朱森强；

【作者基本信息】 合肥工业大学 ， 控制工程（专业学位）， 2025， 硕士

【摘要】 随着自动驾驶技术的快速发展,线控转向作为实现智能驾驶的关键技术受到了广泛关注。相比于传统转向系统,线控转向系统通过电子架构取代机械连接,利用反馈电机提供路感反馈信息,并通过转向电机控制前轮转角,这一创新设计在提升车辆操控性能的同时,也带来了新的技术挑战。线控转向系统的电子化架构显著增加了系统复杂度,使其对各类故障更为敏感。例如,执行器故障、传感器故障、输入迟滞以及状态约束等复杂因素会严重影响系统的稳定性和控制精度。若未能及时检测故障并采取有效的容错措施,可能引发严重的行车安全事故。因此,开展线控转向系统的故障诊断与容错控制研究具有重要的理论价值和工程意义。本文针对线控转向系统在复杂工况下的可靠性问题,展开了故障诊断与容错控制研究,主要研究内容包括:(1)基于键合图理论,建立了线控转向系统的非线性键合图模型,并推导了相应的数学表达式。搭建了线控转向系统实验平台,为后续实验验证奠定了基础。针对线控转向系统的故障诊断问题,设计了一种基于解析冗余关系和故障特征矩阵的故障检测与隔离方法,实现了执行器与传感器故障的快速检测与精确隔离;此外,提出了一种基于滑模观测器的故障估计方法,用以估计故障类型和幅值。(2)针对线控转向系统在传感器与执行器同时故障下的控制难题,提出了一种基于双观测器的神经网络滑模容错控制方法。该方法集成了门控循环单元神经网络和非奇异快速终端滑模控制方法,提高了系统在故障条件下的鲁棒性与可靠性。此外,所设计的双观测器结构能够同时对传感器和执行器故障进行精确估计与补偿,有效提升了控制器在多故障条件下的控制性能。(3)针对线控转向系统在执行器故障与输入迟滞共同影响下的状态受限控制问题,提出了一种自适应模糊固定时间容错控制方法。该方法基于反步控制理论,同时结合模糊逻辑系统、障碍Lyapunov函数与固定时间稳定性理论,在保证系统状态满足安全约束的同时,确保跟踪误差在固定时间内快速收敛,有效地提升了复杂工况下闭环系统的控制精度与可靠性。更多还原

【Abstract】 With the rapid development of autonomous driving technology,steer-by-wire(SBW),as the key technology for achieving intelligent driving,has received extensive attention.Compared with traditional steering systems,SBW system replaces mechanical connections with electronic architectures,utilizes a feedback motor to provide road feel feedback information,and controls the front wheel angle through steering motor.This innovative design not only enhances vehicle handling performance but also brings new technical challenges.The electronic architecture of the SBW system significantly increases system complexity,making it more sensitive to various types of faults.For instance,complex factors such as actuator faults,sensor faults,input hysteresis,and state constraints can severely affect system stability and control accuracy.If faults are not detected in time and effective fault-tolerant measures are not taken,it may lead to serious traffic safety accidents.Therefore,conducting research on fault diagnosis and fault-tolerant control of the SBW system has important theoretical value and engineering significance.This dissertation focuses on the reliability issues of the SBW system under complex operating conditions,and conducts research on fault diagnosis and fault-tolerant control.The main research contents include:(1)Based on bond graph theory,a nonlinear bond graph model of the SBW system is established,and the corresponding mathematical expressions are derived.An experimental platform for the SBW system is established,providing a foundation for subsequent experimental validation.For the fault diagnosis problem of the SBW system,a fault detection and isolation method based on analytical redundancy relations and fault feature matrices is designed,achieving rapid detection and precise isolation of actuator and sensor faults.Additionally,a fault estimation method based on a sliding mode observer is proposed to estimate fault type and amplitude.(2)To address the control challenges of the SBW system under simultaneous sensor and actuator faults,a neural network sliding mode fault-tolerant control strategy based on dual observers is proposed.This method integrates gated recurrent unit neural networks and nonsingular fast terminal sliding mode control methods,enhancing the system’s robustness and reliability under fault conditions.Moreover,the designed dual observer structure can precisely estimate and compensate for sensor and actuator faults simultaneously,effectively improving the control performance of the controller under multiple fault conditions.(3)To solve the state-constrained control problem of the SBW system under the combined effects of actuator fault and input hysteresis,an adaptive fuzzy fixed-time fault-tolerant control strategy is proposed.This method is based on the backstepping control theory and combines fuzzy logic systems,barrier Lyapunov functions,and fixed-time stability theory.It ensures that the system state satisfies safety constraints while ensuring that the tracking error converges rapidly within a fixed time,effectively improving the control accuracy and reliability of the closed-up system under complex operating conditions.更多还原