Prediction Of Shunt Malfunction Of Track Circuit Based On The Combination Of Onboard And Wayside

The jointless track circuit is one of the fundamental monitoring equipments of train control system which has been widely used at home and abroad. It is a typical safety critical system and the key issue affecting the efficiency of railway transport. However, a potential safety hazard which is called shunt malfunction of track circuit has been existed for a long time. The shunt malfunction of a track circuit is always caused by the poor contact of the track circuit and the wheels of the train which is running on this track circuit. The shunt malfunction of a track circuit may affect the efficiency of railway transport and even threats passengers’lives. This potential safety hazard has been the common problem which has to be faced by the train control related workers all around the world.In this paper, for the reason that an automatic and timely detection method is lacked while the researches on characters of shunt malfunction of the track circuit are not yet enough, following works have been done:(1)The ZPW-2000A jointless track circuit was used as the study object. An equivalent circuit model of an occupied track circuit was established based on two-port network theory, the correctness of the model was validated using experiment data based on track circuit physical mode.(2)The electric current signal received by TCR was simulated by the equivalent circuit model of the occupied track circuit.(3) To deeply research in the characteristics of shunt malfunction of track circuit, the related signals received by the monitoring system and TCR were analyzed.(4)According to the characteristics of shunt malfunction of the track circuit, multi-source information fusion theory was used to design a prediction method of shunt malfunction of track circuit. Two kinds of situation were discussed separately:one was defined as a partial or instant situation and the other one was defined as an entire situation. To deal with the first situation, the wavelet decomposition and reconstruction algorithm should be used to extract the features of the current signal received by the TCR. Those features were the input data of the prediction model which was based on the particle swarm optimization algorithm and the support vector machine algorithm.(5)To deal with the second situation, a soft sensor technology which is called Support Vector Machine Regression was used to calculate the value of shunt resistance. The innovations of the thesis are as follows:(1) The onboard information and wayside information are combined to predict the shunt malfunction, which can make up for the past way of monitoring the track circuit.(2) The support vector machine (SVM) classification algorithm was used in the prediction of local (instantaneous) shunt malfunction, and the particle swarm algorithm was used to optimize the parameters of the model.(3) The support vector regression (SVR) algorithm was used in the prediction of section shunt malfunction.