Fault Diagnosis of Brake Train based on Multi-Source Information Fusion

In this paper, the pure air emergency braking mechanism of high-speed train is analyzed, and the emergency braking model of high-speed train is established. Considering that the actual train operation monitoring data is interfered by external complex factors, the fusion filtering of train multi-source braking monitoring data based on Federal Kalman filter is constructed. Based on the recursive expectation maximization identification, the data fusion results are analyzed, and the gradient descent method is selected to optimize the mathematical expectation. The identification results of the brake disc friction coefficient are obtained, and the diagnosis results and maintenance suggestions are given. The simulation results show that the multi-source monitoring data of the train can be fused effectively and the different levels of faults of brake disc can be diagnosed accurately, the effectiveness of the proposed method is verified.