Research On Health Management Of Facilities And Equipment In Power Supply System Of Urban Rail Transit

The stability of urban rail transit power supply system is the basic guarantee for the safe,punctual and efficient operation of rail transit.The operation performance and status of power supply system facilities and equipment are quite different.With the increase of operation mileage,the problems of insufficient maintenance and excessive maintenance are gradually exposed in operation and maintenance decision-making.In the aspects of operation status judgment,failure handling and maintenance mode,it is urgent to make in-depth study on the health management of facilities and equipment.For the health management of power supply system facilities and equipment in urban rail transit,there is a lack of comprehensive health management standards.The research method of health state assessment theory is relatively simple,and fault features are extracted and analyzed by manually exporting data.This affects the analysis of the dynamic trend of the health status of facilities and equipment,and reduces the analysis efficiency and judgment accuracy.This paper analyzes the equipment tree of the power supply system,and studies the data acquisition requirements,hardware architecture and software architecture required for building the health status model of facilities and equipment.Based on the characteristics and indicators of different types of facilities and equipment,a multidimensional and differentiated health management model is formed for power supply system facilities and equipment.The paper builds health management models for power supply system facilities and equipment based on different algorithms such as decision tree theory,KNN algorithm,and logistic regression algorithm.Conduct targeted data cleaning on datasets of different types of devices and import them into the model for training.Conduct multi-dimensional health status evaluation and analysis on devices from various aspects such as the importance of status indicators,health level judgment,parameter distribution,and correlation with health level.The paper also calculates the importance of each state index of the equipment through AHP algorithm and decision tree model,and verifies each other.The paper also predicts the health status trend and health lifespan of devices based on linear regression algorithms.Finally,through the integration algorithm,the voting aggregation model of random forest and gradient lifting tree is used to improve the health management model of power system facilities.The accuracy rate of this model is 99.92% for transformer substation equipment and 97.32% for contact network equipment,which is significantly higher than that of single algorithm model.This paper uses machine learning to extract and analyze the fault feature data of equipment.Compared with traditional methods of health management evaluation,the evaluation methods and results are more perfect,the matching degree of health status of equipment is higher,and the analysis efficiency and judgment accuracy of the trend of health status of facilities and equipment are higher.This paper combines the training results of the algorithm model based on machine learning with the equipment operation and maintenance in the actual operation,carries out multidimensional health status analysis for various facilities and equipment,and formulates the differentiated index scoring standards,equipment rating standards and maintenance measures strategies.The paper improves the health management standards for power supply system facilities and equipment,and provides theoretical support for the formulation of evaluation criteria for operation facilities and equipment of urban rail transit power supply system in actual operation,as well as differential maintenance.