Research On 25Hz Phase-sensitive Track Circuit Fault Diagnosis Method Based On LightGBM

As an important basic equipment in the field of railway signal control in China,25 Hz phase-sensitive track circuit plays an important role in the occupancy inspection of trains in the station.At present,the station mainly uses the Centralized Signal Monitoring System(CSM)to centrally monitor the track voltage signal and phase angle signal of the track relay to realize fault diagnosis.However,due to the relatively complicated environment of the track circuit in the station,the existing two-point monitoring plan of CSM cannot identify certain failure modes of the 25 Hz phase-sensitive track circuit.Therefore,optimizing CSM’s monitoring plan and studying high-performance fault diagnosis method have become a very important and urgent task at present.Therefore,the following research works have been carried out in this paper:1.A monitoring plan is developed,and the signal at each monitoring point of the25 Hz phase-sensitive track circuit is modeled based on the transmission line theory.First,based on the working principle of the 25 Hz phase-sensitive track circuit and the transmission line theory,an equivalent model of the transmission characteristics of the components of 25 Hz phase-sensitive track circuit is established.Then,based on the monitoring of the CSM on the 25 Hz phase-sensitive track circuit,the current monitoring method is proposed to add all feasible indoor current monitoring points to the original CSM monitoring plan,and the monitoring plan of this paper is constructed.Finally,based on the constructed monitoring plan,the information transmission model of each monitoring point is established and verified.2.A 25 Hz phase-sensitive track circuit fault diagnosis model is proposed based on the GA-LightGBM algorithm.First,based on the LightGBM algorithm,a fault diagnosis model with the input features of each monitoring point in the monitoring plan is constructed.Then,combined with Genetic Algorithm(GA),the parameters of the fault diagnosis model are optimized.Finally,a fault diagnosis model based on the GA-LightGBM algorithm is constructed,which will provide support for the design of subsequent fault diagnosis method.3.A 25 Hz phase-sensitive track circuit fault diagnosis method is designed and verified.First,based on the fault injection technology,the monitoring signals under the common failure modes are simulated,and the common failure dataset is constructed and divided.Next,32 monitoring subset plans of the monitoring plan are used as the input feature sets to generate the corresponding common fault diagnosis models based on the GA-LightGBM algorithm,and the monitoring plan is optimized based on the performance of these models.Finally,a fault diagnosis strategy including the diagnosis threshold and model retraining process is designed,and the function and performance of the fault diagnosis method are verified.4.A 25 Hz phase sensitive track circuit fault diagnosis software is designed and implemented.Based on the proposed 25 Hz phase-sensitive track circuit fault diagnosis method and on-site requirements,a fault diagnosis software including online diagnosis,offline diagnosis and model training function is designed.Experiments show that the CSM monitoring plan for 25 Hz phase-sensitive track circuit is optimized.The 25 Hz phase-sensitive track circuit fault diagnosis method based on the GA-LightGBM algorithm proposed in this paper can further improve the fault diagnosis performance of CSM,which can meet the application requirements of the railway field.The amount of figures is 41,the amount of tables is 5,and the amount of references is 58.