基于自适应阈值与扩张状态滑模观测器的AUV执行机构故障检测与估计  

Adaptive threshold and extended state sliding mode observer based actuator fault detection and estimation for AUV

作　　者：吴云凯[1] 胡大海 朱志宇[1] 曾庆军[1] WU Yun-kai;HU Da-hai;ZHU Zhi-yu;ZENG Qing-jun(College of Automation,Jiangsu University of Science and Technology,Zhenjiang Jiangsu 212100,China)

机构地区：[1]江苏科技大学自动化学院,江苏镇江212100

出　　处：《控制理论与应用》2023年第7期1216-1223,共8页Control Theory & Applications

基　　金：国家自然科学基金项目(62173164,62203192);江苏省自然科学基金项目(BK20201451);中国–北马其顿科技合作委员会第6届例会人员交流项目(6–3)资助。

摘　　要：作为自主式水下机器人(AUV)的重要组成部分,执行机构的可靠性对系统的安全运行具有重要意义.本文以AUV六自由度模型为基础,提出了一种基于自适应阈值与扩张状态滑模观测器相结合的故障检测与估计机制.首先,本文将模型中除去控制输入的部分扩张成新的系统状态,得到估计值和实际值之间的残差;其次,针对执行机构的未知扰动,文章设计了一种改进的自适应阈值以监测残差的变化,进一步降低了误诊率与漏诊率;随后,文章在扩张状态的结构基础上设计滑模观测器,将观测器的增益求解转化为线性矩阵不等式(LMI)约束优化问题;最后,通过动态滑模面的设计实现了抖振的抑制并论证了该动态滑模面的收敛性,同时引入等效控制输出误差注入原理,实现了AUV执行机构的故障重构.仿真结果表明,本文所提方法对AUV执行机构的故障具有较好的检测灵敏度和估计精度.As an important part of autonomous underwater vehicle(AUV),the actuator reliability is of great significance to the system security.On the basis of AUV model with 6 degrees of freedom,a fault detection and estimation method via adaptive threshold and extended state sliding mode observer(SMO)is proposed in this paper.Firstly,new expanded state variables are achieved to obtain the diagnosis residuals between actual and estimated outputs.Then,an improved adaptive threshold is designed to monitor the change of residuals,aiming at further reducing the fault mis-diagnosis.Further,the gain solution is transformed into the linear matrix inequality(LMI)constrained optimization problem based on the extended-state structure.Finally,by designing dynamic sliding surface,the chattering phenomenon is suppressed and the convergence can be guaranteed.The fault estimation of AUV actuators is realized via the equivalent control output error injection principle at the same time.The simulation results show that the proposed scheme has an outstanding detection sensitivity and estimation accuracy for AUV actuators.

关 键 词：自主式水下机器人 执行机构 扩张状态 动态滑模面 故障检测与估计 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]