Fault diagnosis of train-ground wireless communication unit based on Fuzzy Neural Network

Abstract

For making up the deficiency of fault diagnosis method of train-ground wireless communication unit of communication based train control (CBTC), a global fault diagnosis method based on model building is introduced. Global fault diagnosis model is mainly comprised of fault symptom, fault diagnosis rule and fault types. Fault symptom is seemed as input space and fault types are seemed as output space. Fault diagnosis rules have two sorts that can interconvert with each other. Firstly, single fault diagnosis rule was designed by fuzzy neural network. Secondly, regarding numerical value variation as vector characteristic, global fault diagnosis rule was designed depending on increasing values range transformation equation, decision matrix and logic algorithm. Lastly, applying global fault diagnosis model and Reworks software development language, fault diagnosis system of train-ground wireless communication (TGWC) unit was realized and correlating experiment validation was carried out. The experiment result indicates that global fault diagnosis model can provide high diagnosis precision of different fault types, had better to analyze cause and effect between fault symptom and fault types, avoid localization to single fault diagnosis method and collectivity precision is 93%.