The State Estimation Method Of High Speed Train Based On Copula

High-speed railway technology in China has made rapid progress under the ambiguous support of the country as well as the researchers, and the train speed has been promoted rapidly for several times. During the long-term service, key components of train bogies will be abrased, resulting in different degrees of degeneration of the component performance. The degeneration will seriously affect the degree of safety and comfort of the operation system. Therefore, the state estimation of train performance is crucial to the safety warning and health maintenance. When the train is in rapid operation, sensors distributed in different locations will monitor mass real-time data. Extracting fault features to estimate the healthy operation of high-speed train through these data is a major and difficult task of present research. According to the non-Gaussian distribution of high-speed train signal, generalized Gaussian model was introduced to fit the distribution of signals. And several new feature extraction methods were proposed in the paper. Several important factors affecting the safety operation of high-speed trains were analyzed. The failure of three key bogie elastic components was researched in the paper. Different degeneration extent of damper parameter and failure of damper in different locations were also researched in the paper.(1) The Copula function theory and its parameter estimation were researched. The features of different types of Copula function and the accuracy and efficiency of parameter estimation of different types were compared in the paper, which paved the way for the selection of Copula model and method of parameter estimation.(2) EEMD transform theory was researched, and the extraction method combining the Copula function and EEMD was proposed in the paper. The joint distribution of IMFs transformed through EEMD was computed using Copula function, and then feature was extracted, which made up the shortcomings of traditional decomposition analysis of every separate component.(3) The evolution law of damper during its parameter degeneration was researched, and the degeneration rate estimation method of high-speed train elastic components based on Copula function was proposed in the paper. Estimation error between the results of proposed method and the real degeneration rate was computed in accordance with the simulation experiment data, which verified the effectiveness of the proposed method.(4) Correlation between different channel signals was studied, and the joint feature extraction method of high-speed train bogie signals using Copula function was proposed in the paper. Mixed fault, single component fault of the same damper in different locations and single component fault of different dampers were analyzed with the proposed method. The marginal distribution parameters, the mean and variance of joint probability density function and some other features were extracted to comprise the feature vectors. The faults were classified by using SVM.