Font Size: a A A

The Correlation Analysis Of The Catenary’s Detecting Data

Posted on:2013-09-20Degree:MasterType:Thesis
Country:ChinaCandidate:Z ZhengFull Text:PDF
GTID:2232330371996213Subject:Control Engineering
Abstract/Summary:PDF Full Text Request
The catenary is one of the most important equipment of the electric railway. Because of the special working environment, the failure of the catenary accounts for a large proportion of the electric railway failures. The pantograph-catenary system works normally is an important premise of the regularly working of the railway. So the detection of the catenary becomes more and more crucial.And the correlation analysis can help to find out which parameters have great influence in some important parameters, these are both important tools to evaluate the working status of the catenary.The data used in this paper comes from XX Railway Administration. As the height and the stagger are the two most important geometric parameters of the catenary, use the Partial Least Square, the Factor Analysis and the statistical analysis methods to analyze the correlation among variables. The PLS is used to analyze different combinations of variables, we get regression equations and display the weight of variables in affecting the height and the stagger with graphics. We can know which variable plays the main role in affecting the two important parameters. Based on the stastistical distribution of the height and the stagger, put forward an index named Direction Quality(DQ) to evaluate the working status of the catenary and the quality of the railway. It’s proved to be suitable for the detecting data.Besides, using Fuzzy Cluster to classify the data according to the similarity of parameters and modeling the result of every cluster with the Support Vector Regression. The models is used to predict the variation range of the height and the stagger, discover the outliers hidden in the data. Also, we can use the models to make evaluations of the catenary’s working status. Results of simulation prove that the classification and the modeling of clusters have a nice accuracy.At last, build the interface that integrate all the methods used in this paper using MATLAB GUI. The interface is supposed to display graphics of different methods, it’s convenient for users.
Keywords/Search Tags:the catenary detecting, the correlation analysis, the Support Vector Regression, the Partial Least Square, the Factor Analysis
PDF Full Text Request
Related items