Fault Diagnosis of Subway Sliding Plug Door Based on Machine Learning and Motor Current Signal

With the rapid development of rail transit in our country, the subway has become one of the most important methods. As a guarantee for passenger safety, the reliability and safety of subway sliding doors have become a hot research object. However, due to the high frequency of use, long maintenance intervals, and complex structure, the probability of failure is high, and the safety of passengers and the normal operation of the vehicle cannot be guaranteed. The failure analysis and fault judgment of the sliding plug door of the system have important reference significance for the normal operation, debugging, and maintenance of the sliding plug door. In this paper, the data processing methods based on feature value extraction and the random forest algorithm are mainly used to judge the two types of faults “down-range pin” and “press wheel”. Compared with the traditional fault tree model, the use of eigenvalues in this study effectively reduces the amount of data processing and emphasizes the differences between different data groups to facilitate subsequent fault classification. And this random forest model can handle a large number of high-dimensional features, making the obtained results more accurate and reliable. The case analysis shows that our method can compensate for the one-sidedness of the results obtained due to the singleness of the fault tree model logic, and provide a reference for achieving high-accuracy fault judgment.