Research On Fault Location Method Of Distribution Network Based On Multi-source Information Fusion

The distribution network is closely connected and closely related to users.Most power users' power outages are caused by distribution network failures.Therefore,the rapid and accurate location of distribution network failures is important for reducing the power outage time and ensuring the safe and stable operation of the power system.However,the weak fault signals in the distribution network,multi-branch,short line length,and a large number of distributed power access have added difficulties for the distribution network location.With the higher and higher degree of automation of distribution network,there is an urgent need for efficient,reliable and intelligent information processing technology to ensure the safe and stable operation and development of the distribution network.Therefore,this paper proposes an efficient and rapid multi-source information fusion distribution network fault location method for collecting,analyzing and processing power grid feature information.First,combined with the law of the amplitude and polarity distribution of the initial instantaneous zero-sequence current of the single-phase grounding fault of the resonant grounding distribution network,a method of fault line selection for distribution network using the current instantaneous value acquisition function of DPMU is proposed.The method can be flexibly adapted to different fault locations,transition resistances,and initial phase angles of the fault;at the same time,based on the multi-source information collected by the main measurement unit DPMU of WAMS,a series of effective multi-source fault feature quantities,such as short window zero sequence current polarity,zero sequence current mutation energy,fault initial phase angle,and switch fault information,can be used as effective information for fault section judgment for subsequent training and analysis.Secondly,on the basis of the multi-source fault feature quantity mentioned above,a fault segment location method based on multi-source information fusion is proposed.Through analysis of common clustering methods,find the PCA suitable for complex distribution network fault analysis,and use it to divide the suspicious fault section,and apply the multi-source information collected by WAMS to the improved Bayesian method.At the same time,the state quantity,electrical quantity and switching quantity are used toobtain the prior probability and the posterior probability,and finally the failure probability of each component is calculated to obtain the fault section.The method has high accuracy,large fault tolerance rate.The multi-source information can make the wide-area measurement system still applicable in the case of missing or wrong transmission.Finally,starting from the Bergeron law of voltage traveling wave distribution along the line,it is proposed that the single-ended voltage traveling wave distribution along the line is unique in the time-distance two-dimensional scale when the single-phase grounding fault of the distribution network,the measured voltage distribution along the line can reflect the only true fault distance;therefore,a post-test simulation method for distribution network fault location focusing on circuit response solution is proposed.The post-test simulation method is composed of a post-test analog waveform library establishment method and a waveform matching method based on time-distance two-dimensional information.Based on the complete post-test analog waveform library,by matching the waveform between the measured waveform and the simulated waveform after the measurement,a value close to the actual fault distance can be obtained.It has been verified that under different fault locations,different transition resistances,and different fault initial phase angles,the location error of the post-test simulation method is within a reasonable and accurate range.