Deep Neural Network Based Suspension Control Failure Identification of Maglev Systems

The comfort and stability of maglev systems are crucial to passenger safety and riding experience. Among them, the suspension gap between the bogie and the rail is one of the critical factors that determine comfort and stability. However, the structural disturbance may cause unstable control of the suspension gap. To address this problem, this study aims to analyze and identify the suspension control failure in a deep neural network approach. An array of accelerometers is installed on the rail to capture the structural response caused by suspension control failure. With the collected acceleration data, a neural network-based classification algorithm is proposed to distinguish the normal and failed suspension control status. As a result, the feature pattern between the normal and failed suspension control can be found in the collected acceleration data, and the proposed algorithm can reasonably identify the suspension control failure. Therefore, this study will help improve the failure identification of maglev systems, as well as the reliability and safety of maglev systems and the comfortable ride experience for passengers.