Study On Non-Strict Input Of Safety Computer Based On Fuzzy Theory

With the rapid urbanized development of China,the railway network is constantly growing,and the passenger flow is continuously increasing,which puts higher requirements on train control system.As the core of train control system,safety computer has a more complex platform structure and application logic.One of the most important developing tendency of safety computer is the increase of the amount of data that requires to be processed,which causes the problem of data inconsistency,thereby brings great challenges to the safe and efficient operation of train control system.The current data comparison structure of safety computer cannot solve the problem of data inconsistency,and cannot meet the requirements of future development.This paper studies data processing problems and architecture improvements of safety computer from the perspective of data layering,structure minimization,and function maximization.The main work of this paper is described as follows:(1)The research status of safety computer is analyzed,it is determined that improving data processing capabilities is the key research object of this paper.Then,problems of data inconsistency and non-strict problems of traditional safety computer are analyzed.According to data characteristics and operating characteristics of safety computer,the data is classified,and the operation data that can cause non-strict problems is determined as the main research object.(2)In order to improve the system's data processing capability,fuzzy mathematics is introduced into the data processing of safety computers according to the application experience of the fields of nuclear power and aerospace.Then,multidimensional feature parameters of data are discussed.In order to improve the efficiency of the system,principal component analysis(PCA)is used to reduce the dimension of the multi-dimensional characteristic parameters.Then,an improved real-time FID3(Fuzzy ID3)algorithm based on historical data and a fuzzy weighted fusion algorithm are proposed to meet the real-time and heterogeneity of train control system.According to the principle of maximizing function and minimizing structure,an improved safety computer data processing architecture is proposed,which combines the traditional safety computer and fuzzy processing model.The fuzzy processing model is consist of two parts,a fuzzy decision sub-model and a fuzzy weighted fusion sub-model.(3)The fuzzy decision sub-model and the fuzzy weighted fusion sub-model are established by Matlab,and models are simulated and tested based on two kinds of practical train operation data.The fault injection and algorithm comparison methods are used to verify the effectiveness and fault tolerance of the fuzzy algorithms.A safety computer data processing simulation platform is built by C#,which simulates the data processing and system operation flow of safety computer.The fault injection simulation results of the fuzzy decision sub-model show that the accuracy of the average single channel data is increased by 11.1%.For injection failures,the average rejection rate is 96.5%.The fuzzy weighted fusion sub-model can correctly output data with an erroneous input or even multiple erroneous inputs,which greatly improves the robustness of the system.The results of algorithm comparison show that the fuzzy algorithms have higher accuracy and fault tolerance in the case of large data fluctuation.The simulation of safety computer data processing simulation platform shows that the data processing architecture can effectively improve the non-strict problems of traditional safety computers,improve the data processing capability and system efficiency of safety computers.Figure 42,Table 9,67 References.