Model-Based Fault Detection and Isolation Scheme for a Rudder Servo System

Abstract

This paper offers a model-based fault detection and isolation (FDI) scheme for a rudder servo system (RSS) that manages ship navigation. Inherent nonlinearities and unknown external disturbances present great challenges in applying FDI technologies to RSS in practice. This paper presents the derivation of the state equations and commonly encountered fault types of RSS from physical laws, and the design of five nonlinear unknown input observers (NUIOs) capable of eliminating the influences of unknown disturbance while detecting and isolating faults. In the NUIOs, no boundary assumption is made on the disturbance, and their parameter matrices are obtained using a linear-matrix-inequality method. Along with an algorithm that programs the logic rules for FDI, the model-based FDI scheme has been implemented on an actual RSS test rig for validating the RSS model and experimentally evaluating it in a real-world environment; both actuator and sensor faults are considered. Based on experimental tests, an adaptive threshold method is introduced to improve decision making and hence effectively eliminate false alarms. Experimental results show that the model-based scheme is efficient and can be used for online FDI.