Feature Extraction of Incipient Fault of Axlebox Spring of High-speed Train Based on CEEMD Sample Entropy

In order to monitor the working condition of the key components of the suspension system of high-speed trains, this paper adopts the method of extracting signal features by combining the complete aggregated empirical modal decomposition (CEEMD) and the sample entropy (SE) information measurement theory. Taking the vibration acceleration of the axle box spring, a key component of the suspension system, under three kinds of non-degree working conditions of 25%, 50%, and 75% of performance degradation as the research object, the vibration signals of the train running at 200km/h, 220km/k, and 240km/h are subjected to the CEEMD decomposition, and a series of simple intrinsic modal functions of the winds are obtained, and the entropy value of the samples is calculated respectively to form a high-dimensional feature vector, then the features are extracted by using a support vector machine (SVM) and sample entropy (SE) information measurement theory. Support Vector Machine (SVM) is used to classify and recognize the early faults. The experimental results show that, as the vehicle running speed becomes higher, the recognition rate is also high, indicating that the characteristics of early failure are obvious, which further verifies the necessity and feasibility of early fault diagnosis of high-speed train suspension system.