Research On Failure Prediction Of 12kV Switchgear Based On Sound Analysis

Generally,the environment in which switchgear is located is more complicated,and various reasons may cause the device to malfunction and affect the normal operation of the device.Therefore,it is particularly important to accurately evaluate the operating state of the switchgear,establish a reasonable evaluation model,and make corresponding health predictions.At present,the operation status of switchgear is mainly obtained through sensors to obtain the relevant voltage and current data,and the operation status of switchgear is judged through the monitoring platform.In this paper,an accurate evaluation method for circuit breaker operation is established by analyzing the sound generated by circuit breaker operation.Through acquisition circuit breaker acoustic signals under normal and abnormal running state,with the improved spectral subtraction for voice signal is collected by noise reduction processing,using the silent period in the rid of the voice signal endpoint detection,to block the sound signal at the same time,to extract the effective sound signal,and the effective voice signal preprocessing.According to the characteristics of the collected sound signal itself,it is analyzed from three aspects of time domain,frequency domain and time-frequency domain respectively,and the index that can reflect the circuit breaker fading performance is extracted.At present,the operation status of switchgear is mainly obtained through sensors to obtain the relevant voltage and current data,and the operation status of switchgear is judged through the monitoring platform.In this paper,an accurate evaluation method for circuit breaker operation is established by analyzing the sound generated by circuit breaker operation.Through acquisition circuit breaker acoustic signals under normal and abnormal running state,with the improved spectral subtraction for voice signal is collected by noise reduction processing,using the silent period in the rid of the voice signal endpoint detection,to block the sound signal at the same time,to extract the effective sound signal,and the effective voice signal preprocessing.According to the characteristics of thecollected sound signal itself,it is analyzed from three aspects of time domain,frequency domain and time-frequency domain respectively,and the index that can reflect the circuit breaker fading performance is extracted.In order to accurately predict the health status of circuit breakers,this paper studies three different short-term data trend forecasting methods,which are grey forecasting method,time series forecasting method and neural network forecasting method,to predict the short-term operation data of circuit breakers.In order to overcome the sharp decline of prediction accuracy after multiple rolling,this paper adopts the combination of rolling prediction and multi-step prediction to forecast the short-term data trend of circuit breakers.By comparing the three prediction methods,it is concluded that grey prediction method and time series prediction method are more suitable for short-term data trend prediction of circuit breakers,while neural network prediction method is more dominant in long-term data trend prediction of circuit breakers.Based on the curve fitting method,studied the variety of long-term data trend forecasting method to predict the breaker long-term operating data,and will adopt a long-term data trend prediction method comparison of the predicted results obtained in the practical application should according to actual condition to select the appropriate long-term data trend prediction method to predict the running status of the circuit breaker.Based on the short-term and long-term data trend forecasting method for medium voltage switchgear,the function relationship between fault characteristics and operation times of circuit breaker is obtained by combining the predicted value and some historical data,and then the running state of circuit breaker is evaluated in the form of health index grade,so as to provide guidance for the realization of state maintenance of circuit breaker.Finally,Matlab software is used to realize the GUI interface design of the failure prediction method discussed above,and realize the human-computer interaction interface to achieve the purpose of visualization,thereby making it more convenient for users to use.