Sensor Failure Diagnosis System Using On Line Refreshable Neural Network
Huang Shan · 2000
This paper presents a sensor failure diagnosis system which can detect, isolate and accommodate the sensors using on line refreshable neural networks for large machinery equipment with its parameters varying or uncertain. The diagnosis system consists of a main neural network and N decentralized neural networks, in which N is the number of non redundant sensors fixed in the large machinery equipment. This sensor failure diagnosis system can also work while the machinery equipment is in dynamic state. The results of digital simulation show that this diagnosis system is very effective.