Research On Circuit Breaker Fault Diagnosis Method Based On Fuzzy Clustering

The circuit breaker is an important device for control and protection in the power system.As the acceleration of the construction of smart grid,the requirement for reliability and stability of the circuit breaker a re also increased.The reliability of the circuit breaker depends on its mechanical characteristics.The fault diagnosis of circuit breaker is often used as input vector for identification methods such as expert system and artificial neural network.The research methods of signal characteristics are even more diverse.Some of these methods require more prior knowledge of the system,and some require precise mathematical analytical models.These conditions are difficult to obtain for typical failures such as the refusal of the circuit breaker.And the general monitoring method is difficult to detect and judge.System operating mode information can be obtained from the abundant monitoring data of the system.Therefore,the state identification and fault diagnosis of the circuit breaker and the processing method of the circuit breaker test signal are of great significance to the safe and stable operation of the circuit breaker.Fuzzy clustering in the fault diagnosis of complex large system of circuit breaker has a unique advantage.At first,this paper designs the circuit breaker failure feature recognit ion system and uses it as a platform for the experimental analysis.The circuit breaker fault diagnosis method based on fuzzy clustering from the perspective of multi-signal detection such as electrical,acoustic and vibration signal is studied in this paper.Then,dynamic weighted fuzzy clustering new algorithm optimized by the peak density which is suitable for circuit breaker fault diagnosis is studied.And its feasibility,high accuracy and superiority are verified compared with the traditional fuzzy clustering.The difference of three dimensional vibration signal characteristics is analyzed,and the influence of different installation place and direction of vibration sensor on the measurement signal is illustrated.Vibration and sound signal are acquired using the influence of the difference.The time differences between the sound signal and the vibration signal are eliminated through the time-scale counterpoint.The sensitivity of the sample time in the peak density clustering algorithm is fixed.Aiming at the characteristics of non-linear and non-stationary characteristics of vibration and acoustic signal,the local mean decomposit ion and approximation entropy are selected as the method of the acquisition of signal characteristics.The simulation breakdown of ZN65 and ZN63 circuit breaker are simulated,and the research on the diagnosis method of double-clustering intelligent fault of circuit breaker is carried out.The training sample cluster is pre-classified based on KFCM optimized by DPCA to establish the membership mapping between the data sample and the fault type.The lowest risk optimal hyper-plane can be obtained.And the SVM is used to build mult iple trainers to put the test sample into trainer.In the end,the final diagnosis result can be obtained by using the mesh optimization algorithm.Through the intelligent fault diagnosis research of the circuit breaker based on fuzzy clustering,an intelligent fault diagnosis method of the circuit breaker based on double clustering with acoustic-vibration joint is designed.The experiment testing proved that this method can better deal with the fault diagnosis problem of the circuit breaker.And the reliability of the diagnostic results can be greatly improved.It effectively promotes the online fault detection work of the circuit breaker.