在感知系统正常工作情况下,从行为学的角度出发,对UUV推进操纵系统异常进行辨识。根据UUV动力学方程设计滑模观测器,与传感器获得的信息进行实时比较,监测系统是否发生异常;当推进操作系统发生异常时,根据根据传感器实时获得的UUV位姿速度信息和推进操纵系统空间分布特点,从辨识行为库中匹配特定的辨识行为,实现对异常点的准确定位,再按照设计的模糊规则得到推进操纵系统异常点的危害程度,以采取相应的应急措施。最后,结合UUV右主推及卡舵异常情况,对基于行为的推进操纵系统异常辨识、异常点定位及异常级别判定方法进行了仿真验证。
An algorithm of propulsion system fault diagnosis for UUV is proposed in this paper. Firstly,we design sliding mode observer based on the kinetic equation to discriminate the presence of abnormal point in the propulsion system. Secondly,according to the characteristics of UUV propulsion system,we develope abnormal identification tasks and abnormal point location rule to identify outliers. At the same time,we design the fuzzy rules to identify the abnormal level. At last,the method used to identify the abnormal behavior is simulated in case of the right thruster abnormal. The results show that the method above is effective and feasible.