Design Of Partial Discharge Detector Based On AE And TEV Detection

In the power system,the power accidents will result in the loss of a large amount of human and material resources.The high-voltage switchgear is the main electrical equipment of the power system.Once the switch cabinet fails,it can lead to serious power accidents.Faults caused by insulation defects account for more than 60%of switchgear faults,and partial discharge phenomena are closely related to it.Therefore,before a power accident happens,the hidden troubles of the equipment can be discovered by detecting the partial discharge phenomenon of the switch cabinet.Based on the analysis of partial discharge phenomenon and its formation mechanism,this paper proposes a wavelet decomposition-Hilbert transform to analyze the feature extraction algorithm of partial discharge signals.By designing hardware circuits and writing software codes,a partial discharge detector based on ultrasonic and ground wave detection was developed.The main research content is as follows:Firstly,the characteristics of partial discharge signals are analyzed and the algorithm of feature extraction is studied.Based on the results of wavelet transform and Hilbert-Huang transform for partial discharge signal processing,the EMD decomposition in Hilbert-Huang transform is replaced by wavelet transform for the problem of uncontrollable number of Hilbert-Huang transforms.This paper proposed the use of wavelet decomposition-Hilbert transform to feature extraction of the signal.Secondly,according to the signal characteristics and the requirements of the acquisition,the detector hardware is designed.The hardware mainly includes three major parts:sensor module,preprocessing module and main processor module.For ultrasonic and ground wave detection methods,the corresponding sensor is designed to obtain the detection signal.Since the amplitude of the detection signal varies widely,an amplifier that can automatically adjust the gain is designed in the preprocessing module.Taking into account the data processing needs and equipment costs and other comprehensive factors,this paper selects STM32 high-capacity chip as the main processor.Thirdly,based on the actual application and data processing needs,the detector software was designed.The software part mainly designs and implements A/D conversion program,SD card file storage program,LCD liquid crystal display program,background noise pre-detection program and signal analysis and warning program.At last,this paper studies the use of pattern recognition to judge the partial discharge phenomenon of the switch cabinet.According to the features extracted by the wavelet decomposition-Hilbert transform,the neural network is used to classify the signals.The test results show that the marginal spectrum can effectively characterize the signal characteristics;the sample signal passes through the pattern recognition system,the accuracy of the classification result reaches 94%;the perforrnance of the main hardware modules and the overall functions of the overall system in the partial discharge detector are correct and reliable.The instrument can perform partial discharge signal detection on the switchgear in the field environment,achieving the expected design goals.