Research on fault diagnosis of the suspension gap sensor based on PBC-PCA-LSSVM

Abstract The suspension gap sensor is essential for the suspension control system of high-speed maglev trains. Its functional failure can have a significant impact on the stability of the suspension point. In the context of engineering applications, the limited number of samples and the high dimension of diagnostic data present a severe challenge to the real-time and accurate diagnosis of sensor malfunctions. To tackle this issue, this paper introduces a fault diagnosis method for the suspension gap sensor of maglev trains based on the point-biserial correlation (PBC), principal component analysis (PCA), and least squares support vector machine (LSSVM). Initially, the PBC and PCA methods are employed to process the monitoring data to reduce its dimension. Subsequently, the pre-processed data is imported into LSSVM for fault diagnosis, which enables the real-time identification of the fault status of the suspension gap sensor. The experimental results show that the proposed method is effective in accurately identifying the measurement faults of the suspension gap sensor while exhibiting a rapid inference speed in comparison to alternative models.