Research Of Identification Of Internal Overvoltage In Power Distribution Network Based On Dual Tree Complex Wavelet Transform

Medium and low voltage distribution network usually refers to the power network with 35kV and below voltage level.It has wide distribution,complex structure and close connection with users.According to statistics,about 70%of the overvoltage in the power system occurs in the distribution network,and the overvoltage and overcurrent caused by it are the main factors that affect the safe and stable operation of the distribution network.All kinds of short circuit faults are often caused by the insulation damage caused by over-voltage,which has a serious impact on the normal operation of the system.So the study of how to extract the Internal Overvoltage signal characteristic and classification,to ensure the safe operation of the power grid has practical application value.This paper summarizes the research actuality about voltage feature extraction and pattern recognition of internal,in-depth analysis of characteristics of power distribution network internal overvoltage,and set up a distribution system simulation model by using ATP/EMTP,7 kinds of internal voltage waveform data obtained through simulation.In-depth study of the time-frequency analysis method of wavelet analysis methods,including wavelet transform and dual tree complex wavelet transform,and focus on the rejection band dual tree complex wavelet transform aliasing properties By comparing the spectrogram of two methods for decomposing the simulation signal,it is proved that the dual tree complex wavelet transform can suppress the frequency band aliasing in the extraction of the dominant frequency feature of the signal.In view of the problem of large amount of data characteristics of dual tree complex wavelet transform,singular value decomposition(SVD)method is introduced to do singular value decomposition for three-phase voltage signals.By comparing the characteristic values of the singular values of overvoltages,it is proved that the dual tree complex wavelet transform and singular value decomposition(SVD)can extract the Internal Overvoltage characteristics,which can be used to identify the Internal Overvoltage in the distribution network.After extracting the overvoltage feature,the classifier learns the characteristic quantity and classify the overvoltage.The performance of the classifier directly determines the quality of the final classification.The network structure and training principle of BP neural network and Deep belief network in neural network are mainly studied,and how to set up parameters of two neural network in this paper in this paper is introduced.In this paper,the white noise of-10dB and-5dB is added into the software simulation model data,and two sets of Internal Overvoltage data samples are formed,the capability of dual tree complex wavelet transform and wavelet transform for Internal Overvoltage signal processing,BP neural network and deep belief network recognition are verified.The results show that the internal over-voltage signal processing ability of the dual tree complex wavelet transform is better than that of the wavelet transform,and the stability and recognition effect of the deep belief network is better than that of the BP neural network.Finally,the single phase to ground fault of the distribution network physical simulation system is collected by the designed record acquisition device,and the method proposed in this paper is further verified.The discovery rate reached 96.25%in recognition of the deep belief networks,proves that the recognition method based on dual tree complex wavelet transform is feasible in the physical simulation system...