Research On Characteristics Of Series Arc Fault Based On Gray Image

In order to study the characteristics and recognition methods of series arc fault,one kind of series arc fault experimental device is developed in this paper,which can carry out the arc fault experiment under the condition of AC and DC.With the aid of experimental equipment,the experimental conditions such as power type,arc gap,power factor,humidity,amplitude and frequency can be changed in series arc fault experiments,and the current waveform of the fault arc under different conditions can be obtained.Due to the different power types and load types,arc fault have different characteristics.In order to find a arc fault feature extraction method which can be equally effective for AC and DC,a method of arc fault feature extraction based on grayscale gradient co-occurrence matrix is proposed in this paper: 6 layers of wavelet packets are decomposed and reconstructed by DB4 wavelet base for one dimension current sampling signal(1024 points),and 64 frequencies are obtained.The current amplitude waveform of the segment,the reconstructed signal is arranged according to the frequency range,and a 64 x 1024 matrix is formed.After the matrix energy is normalized,the gray value of the gray value is 0~255,and then the gray level based on the Weiner filter based on the image degradation model is used to filter the gray level of the structure,and the gray level of the fault and the normal state are obtained.The common noise in the picture is filtered out,and then the image is sharpened by Laplace,and the characteristics of the arc fault are enhanced.The gray level gradient co-occurrence matrix of the grayscale graph is calculated,and the 15 eigenvalues of the co-occurrence matrix are obtained.Due to the volatility and randomness of the arc fault,not all the characteristics of the arc fault are effective distinctiveness.In order to select the effective arc fault characteristics,a feature selection method based on t test is proposed,and the characteristic of the arc fault under the experimental conditions is obtained.Finally,the quantum rotation gate evolution algorithm is used to optimize the c and g parameters of the SVM model.The optimized model is used to identify the arc fault,which validates the recognition ability of the feature quantity extracted by this method for the arc fault.