Fault Data Diagnosis of Energy Consumption Equipment for Urban Rail Transit Based on ART2 Model

ART2 neural network is a kind of self-organized neural network constructed from adaptive resonance theory. Based on the case-based reasoning technology of ART2 Neural Networks and the efficient data processing of database state, we can tell whether the operating state of the city urban train is normal or not and predict the possible failures of energy-consuming equipment. This paper has presented the learning principle of ART2 Model and its specific algorithm process so as to construct the ART2 testing and diagnosis model of energy-consuming equipment state data. We distinguish the energy-consuming equipment state through the four-tier structure of ART2 Model and conduct matching and sorting experiment on the state date of energy-consuming equipment of urban rail trains based on the data generated from the stimulation experimental platform. Results show that compared with the BP neural networks, RBF neural networks and FB neural networks, ART2 neural networks is more advantageous in recall ratio, precision ratio, rate of convergence, clustering results and other aspects of database processing.