Research On Identification Method Of Internal Overvoltage In Modern Distribution Network

Under the background of continuous development of modern distribution network system,continuous operation mode and intelligent technology,considering the active distribution network of distributed power access,from the perspective of power quality,the identification and classification of internal overvoltage in distribution network After analyzing the mechanism and characteristics of several overvoltages with high probability of occurrence,an internal overvoltage identification method based on time-frequency matrix singular value decomposition and multi-level support vector machine(SVM)is proposed.The zero-sequence voltage signal after overvoltage occurrence is selected as the characteristic signal for analysis and processing.Firstly,the signal is fully decomposed by the variational mode decomposition of the adaptive parameter,and then the frequency domain information is obtained by Hilbert transform.The algorithm uses the frequency as the interval to construct the temporary frequency matrix of the distribution network to describe its time-frequency characteristics.By singular value decomposition of the time-frequency matrix,the distribution parameters of the obtained waveform singular spectrum are extracted according to the statistical theory and combined with the time domain characteristics.As the feature vector,according to the different classification tasks of different SVMs in the constructed multi-level support vector machine classifier,the corresponding feature vectors are respectively input into the SVM optimized by the improved PSO algorithm to realize the automatic overvoltage in the distribution network.Identification.The identification method was simulated by MATLAB.The results show that the identification method has the advantage of high recognition rate,which can effectively identify the overvoltage faults in the distribution network,and can still achieve the identification function well in the distribution network with distributed power supply.At the same time,considering the complexity of the actual operating conditions of the distribution network,for the problem that interference may occur in the signal extraction process,the simulation is verified after adding different degrees of noise to the signal.The results show that the identification method has strong resistance.Interference ability to perform identification tasks under complex external conditions.The thesis has 31 pictures,6 tables and 64 references.