Sensors Fault Detection and Diagnosis Based On Morphology-wavelet Algorithm

Abstract

This paper proposed a novel method to fault detection and diagnosis of sensors using trend analysis of input and output signals related to the sensor itself. Firstly, generalized morphological filter with multi-structure elements is designed to filter the random noise and impulse noise in sensor's input and output signals. And secondly, to effectively extract the incipient fault and abruptly fault characteristic, a wavelet transform was used to decompose and analyze the filtered signals in this paper. Through the multi resolution analysis (MRA), the fault can be located accurately. There typical sensor faults such as fix, gain, bias, drift faults were studied. The simulation results show that this algorithm is capable of locating accurately.