Research On Automatic Analysis Of Interoperability Testing Of CTCS-3 On-board Equipment

In recent years,high-speed railway has achieved a great development in China,which makes safety protection of high-speed trains really important.CTCS-3 level train control system is the key equipment for making trains operating safely and efficiently,which has been deployed widely in high-speed railways with max speed over 300km/h.CTCS-3 level train control system consists of on-board epuipment and ground equipment,the interoperability performance of which will have a strong impact on the ability of cross-line operation.For ensuring on-board and ground equipment of different types can interoperate normally,it is necessary to make interoperation test for train control systems,which is also an important step of high-speed railway testing.According to the principle of "only modifying on-board equipment",the test is always prepared for on-board equipment.Testing result analysis is a conclusive work for the whole test.So far,the analysis work is manually done by testing record,expertise and technical standard.However,some potential problems will probably exist for overdependence on manually analysis,including the inefficiency of analyzing procession,too much demand for tester and some inaccuracy caused by human error.In order to overcome these shortcomings,this paper studies the method of automatic analysis for CTCS-3 level train control system,of which related works are summarized as follow:(1)The requirements and objectives of automatic analysis for the test results are proposed,which include the integrality,accuracy and high efficiency.(2)A strategy was designed for automatic analysis for the result of interoperability test,which includes two steps:first,the feature extraction of testing record data based on expert system;later,the classification and identification for testing result samples of which the interoperability feature has been extracted,which is based on support vector machine.The interoperability problem of equipment under test will be found through the process above.Partical swarm optimization was used for parameter optimization of SVM classifier in this paper and it was proved that partical swarm optimization is more efficient for parameter optimization than grid optimization by experimenting.(3)A software for testing record data processing was developed by using C#programing language.Based on testing cases,specification of train control system and expertise,inspection items of the test and rules of feature expection were formulated by the software,which achieved the function of preliminary analysis of test record,feature extraction of testing record data and making more samples for experiments,etc.(4)Simulation and verification was made for the algorithm of classification and identification for testing result samples based on MATLAB simulation platform.The result of experiments showed that the method proposed in this paper can identify the testing result sample effectively and the analysis conclusion was accurate enough for interoperability test for CTCS-3 level train control system.