| Safety has always been the most important issue in the development of high-speed railway in China."Fault—safety" has always been regarded as an important technical index of railway equipment.The CTCS is an important part of high-speed railway and the important technical equipment to ensure the security and efficiency of high-speed railway.However,due to the irresistibility of the surrounding environment and the complexity of the structure of the CTCS,the train control on-board equipment will fail in the actual operation.At present,the diagnosis and maintenance of the train control on-board equipment mainly rely on man,but the manual detection will have the following disadvantages:(1)Because of the lack of fixed format of fault data record,due to the lack of fixed format of the fault data recorded by the electrical staff,the recorded data is missing or incomplete,resulting in low availability.(2)The one sidedness of fault diagnosis,because the existing diagnosis technology depends on expert experience,has a greater one sidedness,making the results inaccurate.(3)The low efficiency of fault diagnosis,because the field survey data are rely on human,the speed is slow,the diagnosis efficiency is low.In view of the above shortcomings of artificial detection,now an intelligent fault diagnosis method is proposed.The basic idea is to use text mining and other knowledge to extract fault features,establish neural network model,and verify whether the fault diagnosis rate meets the requirements.The main structure of this paper is as follows:(1)This thesis first describes the development of fault diagnosis technology at home and abroad,and introduces the structure and function of CTCS-3 vehicle on-board equipments,compareing several intelligent diagnosis methods proposed at present,and proposes a diagnosis method of on-board train control equipment based on GA-ANFIS model.(2)Collect the fault information table of the train control on-board equipment of the high-speed railway line for one year,we can use the expert knowledge for selecting out the representative fault feature vocabulary,then can calculate feature words of the weight in all texts,finally,use the principal component analysis to redundant data,in order to simplify the model.(3)ANFIS is a kind of fuzzy network.It combines neural network and fuzzy reasoning mechanism to learn.However,because of its own structure,the convergence rate changes slowly.It is easy to fall into a local optimization and unable to determine the fuzzy rules.In order to solve the above shortcomings,use the parallel search ability of GA to optimize the parameters of ANFIS;use the K-means clustering method to determine the fuzzy score in order to better reflect the ambiguity.(4)According to the different models of vehicle on-board equipment,we will analyze the differences and similarities between them,and the GA-ANFIS model is continued to be used.Is the observed fault diagnosis curve consistent.The thesis presents an intelligent fault diagnosis method based on adaptive fuzzy neural network.First of all,the fault information table recorded by the electrical staff is collected.Because the collected data is text data,it can not be recognized by the computer.Based on the idea of data mining,firstly,use the Vector Space Model to preprocess the data and then use the Principal Component Analysis to reduce the fault weight data.Secondly,We use the Genetic Algorithm and K-means in order to improve the fault diagnosis rate and accuracy of the model The adaptive fuzzy neural network is optimized by value clustering;Finally,using the 2019 fault data of a high-speed railway line to test and analyze the model,It can be see that the fault diagnosis rate of the train control vehicle equipment can reach 96% based on optimized ANFIS model.Similarly,the fault diagnosis curve of other types of on-board equipment is close to that of 300 T,which verifies the validity,accuracy and universality of the method. |
