The Feature Analysis Of High-Speed Train Monitoring Data Based On Fractional Fourier Transform

Despite all aspects of the technology of China’s high-speed train has been at the forefront, high-speed train is also facing great opportunities and challenges. The major issues at High-speed train safe operation is getting special focus, so it is particularly important that train precaution warning for achieving health maintenance in the process of long-term service. Usually use the method of installing sensor at each important position of high-speed train to collect the vibration signal, to high-speed train service assess accurately and rapidly. Analysis of the characteristics of the high-speed train monitoring data is an important part of the evaluation of high-speed train service state.This article mainly discusses seven working conditions, gradient Parameters working condition as well as a single shock absorber failure conditions.In this paper, first of all, the fractional Fourier transform method is adopted to high-speed train monitoring data for analysis, because the fractional Fourier transform is a time-frequency analysis method of non-stationary signal processing. Fractional Fourier method transform high-speed train monitoring data from time domain to frequency domain, obtaining the score domain three dimensional figure, calculating maximum value under different order, forming the maximum value curves, calculating eigenvalue aiming at the maximum value curves to constitute three-dimensional characteristic value. At experiment part, the characteristics which are obtained by fractional Fourier transform are taken as the input into the support vector machine, then it classifies seven working conditions of high-speed train and obtain the recognition of all the channels. The seven conditions are four of single working conditions, including normal, all air spring damper failure, all transverse shock absorber, resist sinusoidal vibration damper failure and three hybrid working condition of three kinds of hybrid including all air spring damper failure+all transverse shock absorber, air spring shock absorber all failure+all resist sinusoidal vibration absorber and resist sinusoidal vibration absorber all failure+lateral damper failure, so as to conclude channels which are sensitive to four single working condition. Besides, parameters gradient condition and a single shock absorber failure condition is studied by simulating the three dimensional characteristic value.However, the parameter gradient condition deteriorated can not be found, if only the fractional Fourier method is adopted. To improve it, a method of recognising based on the combination of fractional Fourier transform and image processing analysis has been proposed. At the same time the global optimal threshold algorithm about image segmentation is improved. The high-speed train monitoring data is transformed to three dimensional of the score domain, projecting the bottom of it and forming grey shade image to represent the energy accumulation.And then threshold segmentation is done by the global optimal threshold algorithm of gray shade to highlight its main energy region. Finally calculate the similarity between the energy of different conditions and the normal state. Four of single working conditions and three hybrid working conditions are classified effectively by using this method. And the gradient range of parameters of the parameters gradient working condition is extracted.This work is supported by National Natural Science key Foundation of China.