基于ODDD水下机器人故障诊断方法

近年来数据挖掘技术的快速发展使得利用水下机器人作业过程中积累的大量数据进行故障诊断成为可能。基于数据挖掘的故障诊断技术能够从数据中获取潜在的诊断知识。针对水下机器人推进器系统数据特征,提出一种基于聚类和距离的离群点检测方法(Outlier Detection based on DBSCAN and Distance,ODDD)。首先,对数据进行粗聚类,然后采用剪枝规则进行离群点检测,来实现故障诊断。仿真实验结果表明算法能够实现水下机器人快速有效的故障检测。

Outlier detection algorithm based on ODDD for AUV fault diagnosis

In recent years, the rapid development in data mining technologies makes it possible to use massive archived underwater robot operation data to produce scientific decision for fault diagnosis. The fault diagnosis technology based on data mining can find potential diagnosis knowledge from the data. In view of data characteristics of AUV propeller system, a kind of outliers detection method based on clustering and distance is proposed in this paper. First, the data is roughly clustered, and then outliers are detected by pruning rules. The simulation results show that ODDD (Outlier Detection based on DBSCAN and Distance) algorithm is feasible to achieve rapid and effective fault detection for autonomous underwater vehicle system.