Research On Fault Diagnosis Algorithm Of Computer Interlocking System Based On Bayesian Network

The railway station computer interlocking system is the core subsystem of the railway signal system.It is a set of control systems developed on the basis of computer control technology and safety and reliability theory,to ensure the train transportation efficiency and provide train safety protection function.The structure of the system itself is complex,and the failure forms are various,with a high degree of uncertainty,which requires a relatively high level of fault diagnosis technology.At present,the fault diagnosis for computer interlocking systems,in most cases still relies on the field experience of fault repair personnel.Although the combined microcomputer monitoring technology can improve the actual effect of fault diagnosis to a certain extent,it is far from satisfying the demand for the increasing speed of the train and the decreasing train tracking interval for the normal operation of the signal system.It is of urgent practical significance to develop a new generation of intelligent fault diagnosis technology and system equipment for railway signal systems.Bayesian network,as one of the most effective theoretical models in the field of uncertainty knowledge representation and reasoning in recent years,is introduced into the fault diagnosis of computer interlocking systems,which will certainly help to improve the actual effect of system fault diagnosis.This thesis proposes a fault diagnosis method for computer interlocking system based on Bayesian network,and carries out in-depth theoretical research on it by using the diagnosis example of the sub-categories of faults in the local control circuit of the switch:(1)The hierarchical structure of the computer interlocking system and the hardware and software components are divided.The modular fault classification table of the computer interlocking system is given,and the fault subsystems of the local control circuit of the switch equipment are analyzed in detail;(2)Three methods,namely expert experience,FMEA sheet,and FTA,are proposed to comprehensively acquire the knowledge required for modeling the fault system and determine the probability parameters of the diagnostic network;(3)The CME Bayesian network model is established,and the detailed conversion process from the Bayesian network directed acyclic graph to the CME three-layer Bayesian network structure are given;(4)The fault reasoning mechanism of Bayesian network is established,and the cluster tree propagation reasoning algorithm is studied in detail.The establishment of cluster tree,the detailed steps of reasoning and calculation on the cluster tree and the process of the whole network diagnosis reasoning are given;(5)A circuit fault diagnosis example is used to simulate and verify the proposed algorithm in a software environment.The research results show that the Bayesian network-based computer interlock system fault diagnosis algorithm proposed in this thesis can quickly and accurately locate the fault and find the cause of the fault.It is a more effective fault diagnosis algorithm.For systems such as computer interlocking systems,which have complex faults with a high degree of uncertainty,the proposed algorithm has very good application value.