Research On Recognition Method Of Electric Shock Accident Based On Biological Electric Shock Characteristics

Residual current protection device as a low-voltage power grid in an important security of electricity.Its main function is to reduce the damage caused by the abnormal grounding of electrical appliances and other leakage faults caused by fires,and to prevent human electric shock accidents.However,the current operating results show that the existing residual current protection device cannot provide good protection characteristics in extreme weather,sudden heavy load input or electric shock accidents.And there are false or rejection phenomenon.The main reason is that the existing residual current protection device is set according to the total residual current action,and its value is affected by the phase difference angle between the electric shock current and the normal leakage current,there is a protection dead zone.In view of this,scholars proposed to develop a new residual current protection device based on electric shock current action,one of the key issues is to identify the electric shock accident according to the electric shock characteristics.For this reason,this article focuses on the extraction of electric shock features.The main research contents and conclusions are as follows.(1)Analysis of protection characteristics of residual current protection device.On the basis of analyzing the principle of various residual current protection devices,a mathematical model of the protection characteristics of each protection device is established,and the protection characteristics of various residual current protection devices are analyzed.The research results show that the pulse-operated residual current protection device can solve the problem that the current-operated residual current protection device cannot be closed under extreme weather conditions.Amplitude and phase detection residual current protection device cannot provide ideal protection characteristics for non-phase line electric shock accidents.Currentoperated residual current protection devices can theoretically provide better protection characteristics,but the feature extraction of electric shock accidents and the technology of electric shock current separation are not mature enough.(2)Analysis of frequency spectrum characteristics of electric shock signal.Using the time-frequencyamplitude window provided by the S transform,the frequency characteristics of the residual current signal under the single-phase and three-phase circuit structures are analyzed,and the influence of the electric shock time and the weight of the electric shock organism on the residual current spectrum characteristics is initially discussed.The research results show that the main frequency components of the residual current of the singlephase circuit and the three-phase circuit after electric shock are distributed in the range of 0Hz-500 Hz and0Hz-1000 Hz respectively.In the two circuit structures,the residual current at the time of electric shock has a surge in amplitude of the frequency components of 1000Hz-5000 Hz.The amplitude of the high-frequency component of the residual current at the time of the electric shock at the zero crossing point is weakened.(3)Recognition method of electric shock accident based on wavelet high frequency characteristics and PNN.Extract the wavelet high-frequency distribution of each layer through the multi-scale frequency window provided by the wavelet multi-resolution analysis.And use the normalized processing of each layer of wavelet high-frequency distribution of the accumulation and quantification of the high-frequency characteristics of the first five layers of the residual current signal to describe the electric shock accident.Fully consider the randomness of the time of electric shock accidents,classify the extracted wavelet features,and further build a PNN-based electric shock accident recognition model.And according to the specified step length,the network smoothing parameters in the defined domain are optimized,and mean clustering method is used to optimize the network structure.The research results show that the normalized wavelet highfrequency distribution amplitude change amount of each layer can be better described in the corresponding stage of the wavelet high-frequency distribution amplitude surge phenomenon.The optimal smoothing parameter interval of the established PNN network model is 0.15-0.29,and the corresponding best recognition rate of electric shock accident is 95.5%(4)Recognition method of electric shock accident based on cyclic spectrum characteristics and cluster analysis.Using cyclic power spectrum analysis to obtain the three-dimensional diagram of the cyclic power spectrum of the residual current signal before and after the electric shock and the slice diagram of the main frequency components to extract the characteristics of the electric shock signal’s cyclic spectrum.Further use K-means clustering analysis to cluster the combined features of the cyclic spectrum of different dimensions to extract the electric shock recognition criterion.At the same time,a modified Euclidean distance measure is proposed to improve the accuracy of cluster recognition.The research results show that the combined features of the cyclic spectrum 2,3,and 4 of the single-phase circuit have the highest recognition rate of94.67% compared with other dimensional features.The corresponding cluster centers of residual current before and after electric shock are 20.597,57.682,4.773 and 4.102,11.387,0.923,respectively.The best recognition feature for a three-phase circuit is the cyclic spectrum feature 4 The recognition rate is 100%,and the corresponding cluster centers of residual current before and after electric shock are 16.136 and 2.197 respectively.The revised Euclidean distance measurement has a 99.33% recognition rate for electric shock accidents.(5)Research on Hardware Implementation of Electric Shock Accident Feature Extraction Based on Arduino Due System.Use the Arduino IDE to write the control program for the feature extraction of the electric shock accident cycle spectrum,and compare the feature extraction results with the Matlab software simulation platform extraction results.The research results show that the cyclic spectrum feature extraction of electric shock accidents can be realized on the Arduino Due development board.The calculation result is different from the line spectrum amplitude calculated by Matlab software simulation,but the line spectrum distribution is basically the same and the proposed cyclic spectrum characteristics are still reflected.