Method of Extracting Characteristic Parameters of Medium-speed Maglev Train Levitation Controller based on Relief Algorithm

摘要

Medium-speed maglev train has been widely studied as a new means of transport. The suspension controller is the center of the suspension control system for electromagnetic medium-speed (EMS) maglev trains. Its inner equipment and components are complex and easily damaged. In severe cases, it will directly cause the train to lose its suspension ability. Therefore, efficient and accurate state identification and fault diagnosis algorithms are necessary for suspension controllers. In this paper, the medium and low speed maglev train fault data of Changsha Maglev Airport line is used as the basic data, to analyze the determined eight main faults of suspension controllers for medium-speed maglev trains. These faults including filter failure, single resistance open circuit of charging circuit, double resistance open circuit of charging circuit, protection circuit failure, input power failure, DC / DC converter failure, LC control power supply circuit malfunction, capacitor open circuit of charging circuit. Based on the Relief algorithm, the characteristic parameters of the suspension controller faults are extracted with sufficient correlation. Finally, the characteristic selection results are verified by BP neural network, and the accuracy rate is significantly improved by 9.44% to 88.33 %, which proved the effectiveness of the method.