Research On Feature Extraction Technology Of Distribution Network Operation Fault Based On Multi-source Data Fusion

The development of the new era is extremely rapid,and the degree of automation of the distribution network is also rapidly increasing,which greatly promotes the convenience of obtaining fault data.Therefore,when the distribution network fails,the accident information is large and complex,and it will quickly flow into the dispatch center.In this case,the dispatcher needs to quickly and accurately know the most core alarm information.However,it is very difficult to identify faults quickly and accurately.The main reasons are misjudgment and missed judgment.Therefore,the personnel who process the information need to rely on the effective distribution network fault feature extraction theory and method to provide fault feature information,which can be used as auxiliary evaluation to ensure the safe operation of the distribution network.In the past,a series of switch information generated by similar protection devices and circuit breakers is the most common method for identifying and extracting fault features.After acquiring feature information,the location of the fault is determined by some intelligent algorithm.This requires that the completeness of the switching quantity is extremely high,because the accuracy and completeness of the fault information have a great influence on the result of fault feature extraction and recognition in a certain sense.Therefore,based on this,the paper pays attention to the change of electrical quantity before and after the fault,and comprehensively analyzes the two types of information,and designs a targeted fault feature extraction algorithm.This paper first,introduces the multi-source operational data system closely related to the fault occurrence of distribution network,and proposes a multi-source heterogeneous information model for distribution network for the wide variety of multi-source data sources and the different data composition.After unifying the format of multi-source data,the improved rough set theory and Bayesian network are combined to extract the characteristics of the switch changes such as protectors and circuit breakers,and the wavelet energy analysis is used to extract the characteristics of electric quantity changes before and after the fault occurs.Finally,Bayesian failure degree,wavelet irradiance,wavelet failure degree and wavelet singularity are proposed to characterize the severity of line faults,and the DS evidence theory is adopted to multi-source data fusion of the above evidence,so as to extract fault features.Identification provides a comprehensive reference.Matlab programming and PSCAD simulation also show that compared with the traditional method,the multi-source data fusion based distribution network fault feature extraction method can improve fault feature extraction,diagnostic reliability and accuracy.Can be widely used in the direction of power grid fault diagnosis.