engleski

Fault detection and localization on underwater vehicle propulsion system using principal component analysis

In order to achieve full and viable autonomy of an underwater vehicle, it is of high importance that the vehicle can operate in at least a foreseeable, if not an unhindered way, under conditions of a technical failure or breakdown. Thus, a technique for dealing with a broad range of possible faults and failures is needed. Every such technique can be methodologically separated into a sub-technique for fault detection and an adaptive mechanism on the control law or behavior. This paper concerns itself with the problem area of fault detection and localization. It describes the mathematical method of Principal Component Analysis and provides guidelines for its use on fault detection and localization. The benchmark example used herein is a DC motor as a simplified model of a propulsion subsystem of an AUV. The particular algorithm in the PCA class that will be elaborated upon in this paper is the Partial Principal Component Analysis. The paper gives results of simulation in Simulink, comments on the successes and shortcomings of the method, and contains proposals for future research.

fault detection; fault localization; autonomous underwater vehicle; principal components analysis; PCA

nije evidentirano