Research On Track Circuit Health State Perception Method Based On PHM Technology

As one of the key signal basic equipment,the track circuit plays an important role in ensuring the safe running of the train.When the carrier frequency signal is transmitted to the receiver at the receiving end,it will be quickly attenuated due to the high inductance of the steel rail.In order for the signal to reach the receiver,compensation capacitors are set at equal intervals in the middle of the steel rail.The reduction of the capacitance value of the compensation capacitor or the disconnection will cause the red light belt to fail;the track circuit reader device of the train cannot receive the signal,and the phenomenon of "code dropping" will occur,which endangers driving safety and affects driving efficiency.The existing compensation capacitor detection method in my country is to find faulty compensation capacitors through regular inspections of electrical inspection vehicles.During the inspection interval,the faulty compensation capacitors can not be found,and it is difficult to repair them in time.The railway electric affairs department proposes the transition from "fault repair","timed repair" to "status repair",and compensation capacitor state monitoring is the prerequisite for realizing this change in maintenance.In this thesis,in the shunt state,the influence of the compensation capacitor on the shunt current curve under the state of different capacitance values is studied,and a method of monitoring the state of the compensation capacitor is proposed.The main research work of the thesis is as follows:First,based on the transmission line theory and the two-port network theory,the rails and electrical components in the track circuit are modeled,the two-port network model of the compensation unit is established,and the three situations when the train is split through the compensation capacitor unit are considered to establish the splitting situation Below is the equivalent two-port network model from the transmitter to the train wheelset.Set the corresponding track circuit simulation parameters,change the capacitance value of the compensation capacitor,simulate the data of the shunt current curve under different capacitance values,and summarize the influence of the compensation capacitor failure on the shunt current curve by comparing the curve change trend.Secondly,perform CEEMD decomposition and LMD decomposition on the shunt current curve,calculate the fuzzy entropy value of the components obtained by the two decomposition methods,and combine them into eigenvectors,which are used as the input of the compensation capacitor state monitoring algorithm.The WOA-LSSVM algorithm is proposed to monitor the state of the compensation capacitor,and the whale algorithm is used to find the two key parameters of LSSVM,the penalty factor C and the radial basis kernel function parameter.In view of the fact that the whale algorithm is easy to be trapped in the optimal operation result of a smaller interval,by introducing a random differential mutation strategy,the whale searches for the prey stage towards a larger range and seeks solutions at a longer distance,and jumps out of the smaller interval.Aiming at the problem of slow convergence speed,adaptive weights are introduced and applied in the position update of the bubble network attack stage,which improves the ability of the algorithm to find the optimal solution locally.Use the improved WOA-LSSVM algorithm to monitor the state of the compensation capacitor,with a high recognition rate.Finally,CEEMD decomposition is performed on the shunt current curve to calculate the energy entropy and form the eigenvector.The phase space reconstruction is carried out on the characteristic vector of the shunt current curve,the one-dimensional signal is expanded into the high-dimensional space,the HSMM model in each state of the compensation capacitor is established,and the HSMM model is trained using the reconstructed data.The test sample data is input into the HSMM model of each compensation capacitor state,and the state of the compensation capacitor is judged by outputting the logarithm of the likelihood probability.The simulation shows that the two methods have a higher recognition rate for the state monitoring of the compensation capacitor.