Fault Diagnosis Technology Of High Voltage Circuit Breaker Based On Particle Swarm Optimization Extreme Learning Machine

The circuit breaker bears the task of control and protection in the power system.It is an extremely important electrical equipment in the power system.If the circuit breaker fails,it will cause huge losses.On-line monitoring and fault diagnosis of high-voltage circuit breakers can detect faults in time,provide conditions for periodic maintenance and maintenance of circuit breakers,improve reliability of power supply and distribution,and reduce maintenance costs.At present,there are still some problems in the online monitoring system of high-voltage circuit breakers,which need further research and discussion,such as improper sensor selection and inaccurate parameter measurement;eigenvalue extraction and fault diagnosis methods are not perfect enough to affect the diagnosis results.In this paper,the "VS1-12" vacuum circuit breaker is taken as the research object,and the online monitoring and fault diagnosis of the mechanical characteristics of the circuit breaker is studied.The main research contents are as follows:(1)By setting up the experimental platform,the typical faults such as the winding core of the circuit breaker opening coil,the secondary circuit voltage is high,and the secondary circuit voltage is low are simulated,and the contact stroke signal and the closing coil current signal are collected.,lay the foundation for feature value extraction and fault diagnosis.(2)Based on the contact travel curve,a eigenvalue extraction method based on the derivative(seeking speed)method is proposed,and a Modified Ensemble Empirical Mode Decomposition is proposed for the opening and closing coil current.The eigenvalue extraction method of the MEEMD algorithm is based on the above method,and the characteristic values of the contact stroke curve and the opening and closing current curve of the 120 sets of circuit breakers under normal and fault conditions are extracted and a eigenvalue sample library is created.(3)Established three kinds of circuit breaker fault diagnosis models based on Extreme Learning Machines network,Genetic Algorithm optimized ELM network and Particle Swarm Optimization optimized ELM network.And use two kinds of diagnosis methods of single information and multi-information to carry out simulation analysis of circuit breaker fault diagnosis.(4)Using Java to design the online monitoring and fault diagnosis upper computer software system of the circuit breaker,the system includes several modules such as "state curve display","eigen value extraction" and "fault diagnosis";meanwhile,the MySQL database is used to store data.to achieve the query and traversal of the circuit breaker history information.