Fault Detection And Analysis Method Of Turnout Conversion Equipment Based On CNN-SVM Algorithm

Turnout equipment(switch machine)is a very important link in rail transportation.It undertakes turnout position,reverse position and determines the train’s running direction.In order to ensure complete turnout conversion,avoid deviation from the basic rail,and ensure the safety of the train,it can act in a timely and safe manner.At the same time,it is also the action monitoring device,through the signal component to timely reflect the fault,for the maintenance personnel to quickly complete the fault repair.In this paper,based on the current development of turnout equipment at home and abroad,this paper proposes a convolutional neural network support vector machine(cnn-svm)fault diagnosis model based on dimension reduction of principal component analysis(PCA).Based on domestic and foreign literatures,through the development of simple model,rapidly improves and perfects the model,and formulates practical research contents and research methods.In this paper,the principal component analysis method,convolutional neural network,and classification algorithm are theoretically explained.In the aspect of data dimension,the principal component analysis method is used to reduce the dimension of data and effectively retain the effective information.On the one hand,it reduces the redundancy of data;on the other hand,it ensures that the data is more representational as far as possible.In the aspect of feature extraction,by analyzing the convolutional neural network,it is found that it has the characteristics of spontaneous learning,which avoids the complexity of manual feature extraction in the development of simple model,and effectively improves the fault diagnosis recognition rate of manual feature extraction.In the aspect of classification algorithm,by comparing classification algorithms,it is determined that SVM has a higher fault recognition rate based on the data of this study.After research and development,the cnn-svm fault diagnosis model processed by PCA was finally established,while other classification models were retained to meet the needs of different scenarios.