Multi-Channel Methods Analysis High-Speed Train Data

High-speed train running safety is an important part of the train system. With the rapid development of railway transportation industry in China, just ten years have finished six big speed, railway in our country has gradually entered the characterized by "high-speed passenger and freight overloading" era, with the speed for the entire network and is the higher requirement of railway system of train operation safety. With the increase of train service time, vehicle dynamics behavior will occur a variety of changes, on the train running state judgment becomes even more difficult. This topic through the study of multiple sets of installation on the train bogie sensor data, analysis of train real-time data change rule under some particular condition, and then to evaluate vehicle safety state.In this paper, first of all to give a brief introduction of the sensor data, involved in the experiment of several kinds of shock absorber made in detail and their respective function. Analyses the commonly used several time domain analysis parameters and time-frequency analysis method, is given in this paper, the selected time domain features and the reason and purpose of S transform.With the selected several time domain features of various working condition was analyzed, and the extracted in the time domain characteristics of relatively high degree of recognition and theory value, using support vector machine (SVM) to classify fault condition. In four kinds of single condition and working condition of three kinds of hybrid resolution has achieved a certain effect, and in this paper gives the resolution of the different parts of the channel under different speeds.By using the method of S transform, completed to various speed conditions, the processing of a single condition and mixing conditions all channels, discrete signal transform too S frequency domain transform matrix. Extract the modulus of matrix under the fundamental frequency of various characteristics of input support vector machine (SVM) classification, is given in four kinds of single fault condition and three kinds of mixed fault conditions in the resolution of all the channels.Based on time domain and the effect of S transform its classification, this paper puts forward the concept of compound channel identification, the physical sense channel of interconnected into consideration, the working condition of type common recognition. Attempts to apply the theory to four kinds of single fault condition and three kinds of mixed fault conditions resolution, obtained significantly higher than the single channel alone recognition classification effect, and the recognition rate of these channels are given in this paper. For other high-speed data research method provides a new way of thinking.