Neural Network-Based Method For The Life Prediction Of Computer-Based Interlocking System

Computer-Based Interlocking system(CBI)is the core technology and equipment of railway signal,and it is also an important part of ensuring safety and efficient operation of the train and railway station operations.At present,China's CBI life cycle management approach mainly refers to the measures for the administration of the traditional relay interlocking,but it is lack of analysis and evaluation methods for the life of electronic equipment of the system which limits largely the scientific management of CBI's life.Based on the architecture of the system,this paper studies a method of evaluating the service life of CBI based on neural network.The main work is as follows:(1)After analyzing the current situation of the application of CBI and the research status of system life prediction methods in various fields,this paper puts forward the life prediction scheme which is suitable for CBI,and gives the concrete implementation steps of life prediction;(2)Combining with the analysis of the hardware structure and component function of CBI,the fault tree model of CBI is established by using CBI based Double 2-vote-2 as the main research object.The minimum cut set of the fault tree model is obtained by qualitative analysis to draw a conclusion of the relationship between the component failures and the system failure,and the neural network training data set is constructed.(3)The method of the neural network prediction performance improvement is studied,and the parameters of the central neuron width matrix of GRNN neural network are optimized by Particle Swarm Optimization(PSO)algorithm,which effectively improves the accuracy of neural network prediction.(4)In order to reflect the hardware redundant structure of CBI fully,three different CBI life prediction models were established based on GRNN neural network,the modified GRNN neural network and BP neural network.(5)For three neural network-based models for the life prediction of CBI,the best prediction model can be selected through the network performance comparison and analysis combining with the characteristic of the system.Finally,using the operating data,the AB-typed CBI which is widely used in China's railway,and it is regarded as the object to predict the system life and verify the effectiveness of the prediction model and method.In this paper,through the theoretical analysis and example verification,a neural network-based method for the life prediction of CBI is proposed to realize the scientific prediction of the service life of CBI,which can provide some reference for the operation and management of CBI.