This paper is a detailed study on artificial neural network, fuzzy logic, fuzzy neural network and the theoretical basis when they are used in detection and prediction of sensor fault. And in it an approach of on-line detection and prediction of sensor fault based on the respective advantages of artificial neural network and fuzzy logic is proposed. On the basis of T-S Model, this approach builds fuzzy neural network, employs on-line refreshing network, gives predictive signals and then, detects sensor fault. Judging by the findings from experiments and simulation results, the approach can accurately detect sensor fault, separate the fault sensors and in the meantime, it provide system and equipment with rather high-qualified temporary predictive signals. With further analysis, the approach also possesses no-model feature.
|