NEURAL-NETWORK-BASED SCHEME FOR SENSOR FAILURE-DETECTION, IDENTIFICATION, AND ACCOMMODATION

This paper presents a neural-network-based approach for the problem of sensor failure detection, identification, and accommodation for a fli ght control system without physical: redundancy in the sensors. The ap proach is based on the introduction of on-line learning neural network estimators. For a system with n sensors, a combination of a main neur al network and a set of n decentralized neural networks achieves the d esign goal. The main neural network and the ith decentralized neural n etwork detect and identify a failure of the ith sensor, whereas the ou tput of the ith decentralized neural network accommodates for the fail ure by replacing the signal from the failed ith sensor with its estima te. The on-line learning for these neural network architectures is per formed using the extended back-propagation algorithm. The document des cribes successful simulations of the sensor failure detection, identif ication, and accommodation process following both soft and hard sensor failures. The simulations have shown remarkable capabilities for this neural scheme.