Research On Test Sequence Automated Generation For Interconnection Of New Train Control System

Compared with the traditional train control system,the new train control system(NTCS)reduces track circuits and transponders,and integrates the equipment in trains and stations.Changes have taken place in integrity checking,positioning,blocking formats,operating modes,and wireless communications.In order to ensure the accuracy,security and real-time performance of information exchange between different types of devices,and to promote the improvement of device functions,it is necessary to design test cases for interconnection testing.The automatic generation of test sequences helps to realize test automation and improve test efficiency.This thesis takes the interconnecting equipment of the new train control system as the research object,and solves the problems of large workload,low efficiency,the need to accumulate a lot of expert experience in manually compiling test sequences.A deep learning method is proposed to realize the automatic generation of test sequences,and to judge the correctness of the generated tests.The specific research contents are as follows:(1)Key elements that affect the generation of test sequences are identified and extracted.The architecture,the operation mode,mode conversion,and typical scenarios of the new train control system are analyzed.The interface relationship of train control equipment is analyzed,and the influencing factors of equipment interconnection are determined.Analyzing the test cases and using data-driven testing methods to complete functional testing and R&D testing of equipment.(2)A method for automatic generation of test sequences based on Bert model is proposed.The problem of concatenating test cases is transformed into a textual logic connection problem based on train operation scenarios.Based on the pre-training and data set construction of the Bert model,the debugging of the model is completed through the existing test sequence.Using the cross-entropy loss function and the accuracy function to measure the model performance.The test sequences generated by the model are compared with those written manually to prove the feasibility of the method.(3)A method for judging the correctness of test sequences based on rule engine is proposed.The correctness judgment criterion is defined,and a rule base is formed including the train operation mode conversion,level conversion,test case coverage,etc.Based on the Drools rule engine,the correctness of the test sequence is judged.For test sequences that do not meet the requirements,output the reason.(4)Using C# language to complete the development of the new train control system test sequence auxiliary tool.Completing the detailed description of the operation and phenomenon in the test document,and realizing the electronic management of the test document.Finally,based on the laboratory test platform,the sequence is tested.The test results show that the test sequences are correct and available,and covers all test cases.It can discover the functional defects of the equipment under test and improve the function of the train control system.Figure 38,Table 15,66 References.