Study On Fault Diagnosis Method Of Track Circuit Based On Improved Particle Swarm Algorithm

Track circuit is one of the most important equipment in CTCS (Chinese Train ControlSystem), and the train operational safety is affected by the quality of its signal transmissiondirectly. When a fault occurs on track circuit, it may cause serious safety accidents. At present,the methods for fault diagnosis of track circuit equipment exist some shortages, such asinaceuracy in diagnosis, inefficiency of diagnosis, more rely on the repair work experienceand so on. The traditional fault diagnosis methods for track circuit cann’t meet efficient,intelligent, comprehensive requirements. Therefore, adopt advanced fault diagnosis tehnologiesand build an intelligent fault diagnosis system of track circuit have an important significanceto find hidden danger, analysis fault, improve the maintenance level.This dissertation mainly takes ZPW-2000A track circuit as the research object. Itanalyzes the working principle, technical characteristics, common fault and the fault mechanismof ZPW-2000A track circuit. The detection algorithm of track circuit is improved, and the faultdiagnosis model and database of track circuit is perfected. Thus, the dissertation realizes thedetection and fault diagnosis of track circuit. This dissertation mainly completes the followingcontents:Firstly, the dissertation analyzes the working principle of ZPW-2000A track circuit,and summarizes the common fault of track circuit. According to the transmission line theory,the model of four terminal network is established on track circuit. The compensationcapacitor and the ballast resistance effects the amplitude envelope, which is summarized.Secondly, the dissertation studys deeply the principle, parameter selection, algorithmprocess of the particle swarm algorithm. In order to improve the search ability of the particleswarm algorithm, the particle swarm algorithm is improved. Aiming to the lack of BP neuralnetwork, such as slow convergence speed, easily falling into local optimum, the improvedparticle swarm optimization algorithm is adopted to optimize the parameters of BP neuralnetwork. A neural network modeling method is proposed based on improved particle swarmalgorithm, and the optimization process is given.Finally, the common failure of the ZPW-2000A system is analysis. Using VC++6.0andMatlab joint programming methods makes a human-machine interface for track circuit faultdiagnosis system. According to the field data, the dissertation detectes the track circuit andalso fault diagnosis. From the diagnostic results, we can see this system can diagnoseaccurately the fault of track circuit, which judge fault type and analyse fault reason. All kinds of fault is processd timely, so the diagnosis efficiency is improved.