Fault Detection And Section Location Of Distribution Network Based On Synchronous Waveform Analysis

The distribution network lines have characteristics like complex structure,a large number of types,many branch points and wide coverage,so the distribution network is under poor operating conditions.And short-circuit faults are prone to occur in distribution network lines and most of which are single-phase-to-ground faults.With the construction of smart grid and the continuous access of distributed power to the power system,the distribution network will develop into an active network with dual-direction power flow in the future.Existing fault diagnosis schemes mostly rely on amplitude and phase angle information of voltage and current or component switching states in local partitions,and there is no unified time-stamped synchronous data sources between various sections as well as different places.So,it is hard to integrate multi-source information from a global perspective to achieve more accurate and reliable fault diagnosis and meet the requirement of safe and reliable operation.With the development and promotion of synchronous measurement system,high-precision synchronous measurement devices have further covered important nodes of distribution network.Massive accurate unified time-stamped synchronous information will be collected in the dispatching center.Waveforms of electrical quantities can be recorded in the form of time series,and these waveforms contain abundant information which can reflect the system operating state.The centralized fault diagnosis method based on multi-node measurement data can monitor the system dynamic process under different faults by analyzing the global measurement information formed by data of multiple nodes,which can achieve deeper understanding and stronger control of the overall system state.In this paper,diagnosis of single-phase-to-ground fault in distribution network under the background of the continuous development of high-precision measurement system is studied.Fault features of synchronous waveform recorded by measurement terminals are extracted and analyzed.Two kinds of data-driven single-phase-to-ground fault diagnosis methods are proposed.On this basis,early warning of evolutionary permanent grounding fault is analyzed.The main research contents are as follows:(1)The single-phase-to-ground fault diagnosis method based on mining association rules between synchronous waveform features is studied.Features are extracted from the accumulated historical fault current waveforms and discretized to construct the fault transactional database.Then association rule algorithm is improved to mine the database to obtain a single-phase-to-ground fault diagnosis rule base.Finally,the fault diagnosis result is given by finding the optimal rule(2)The single-phase-to-ground fault section location method based on analyzing topology matrix is studied.The waveform feature set is adjusted and expanded.Waveform features are extracted from historical fault current waveforms to establish a topology matrix.The random matrix theory is introduced to map historical fault samples into the random matrix eigenvalue space.Then obtaining the cluster central point corresponding to each section in this space.Finally,section location result is given by comparing the distance between the point of the new single-phase-to-ground fault and each cluster central points.(3)The early warning method of evolutionary permanent grounding fault is studied.The gradual development law of permanent grounding fault and the principle of transient grounding fault are analyzed.On this basis,the characteristics of evolutionary permanent grounding fault are summarized,and waveform features are extracted from waveforms of each transient grounding fault.A function about insulation deterioration is constructed using these features,which tracks and reflects line insulation deterioration,and provides some guidance for the early warning of permanent grounding fault.