Study Of Integrative Transient Protection And Fault Location Based On Distributed Intelligence For Feeders In Distribution Systems

An integrative transient protection and fault location based on distributed intelligence for feeders in distribution systems are studied in this thesis, which are of the key questions and difficult points existing in field operations of domestic and foreign small current grounded systems. The breakthrough at a series of creative studies has been brought on single-phase-grounded-fault protection, multi-phase-fault current protection, single-phase-grounded fault location and multi-phase fault location based on the principle fusion of wavelet analysis and distributed intelligence. These beneficial studies have been done in order to improve on distribution automation and the reliability of power supply. Distributed laws and associated features between the three phase transient current components (TCC) are analyzed in theory, on the basis of sufficient researches to formed mechanism and distributed features of the single-phase-grounded-fault transient state process in non-effectively grounded systems. Creative studies are as follows: TCC of each phase is regarded as basis foundation of grounded fault protection; Feature band measure of phase TCC based on only one or multi-compositive feature bands is extracted by multi-scale decomposition algorithm of orthogonal wavelet; A new principle of single-phase-grounded-fault protection with theoretical judge criterion is built up based on the theoretical ratio between feature band measures of faulty phase and healthy phase; Another new principle of single-phase-grounded-fault protection is proposed by extracting the relative phase associated features between the phase TCC by combinatorial complex wavelet transform. These protective principles proposed can be used for all kinds of non-effectively grounded distribution systems, and their protective performance has not been influenced on system operation mode and system parameters, has no relation with the load operation performance and asymmetry, fault resistance, fault distance, fault instant angle and so on. Action valve of protection is abandoned without the need for setting calculation. And they can be fused with the gap units of distributed feeder protection, and do without disposing appropriative devices for the selection grounded feeders. The protective speedy act is combined with auto-reclosing mode to greatly enhance the reliability of power supply. Therefore, the protections proposed have excellent performance of higher sensitiveness and reliability, more strong auto-adaptive ability and generality. These fruits of above protection researches in this dissertation have been shown outstanding innovation from overcoming many difficulties, and have important theoretic significance and practical values. Based on foregoing study, a system of feeder fault location for single-phase-grounded fault in distribution systems is proposed by the fusion of neural network and feature band measure vector extracted through decomposition algorithm of orthogonal wavelets packet. Two key problems improving accuracy of fault location have creatively been solved. The one is proposed that healthy phase TCC is taken as basic foundation of single-phase-grounded fault location, which is different from information of fault phase as basic foundation of fault location generally. Another one is that the modified principle of faulty instant angle and its method are proposed by acquiring law of faulty instant angle with fault location. The equivalent algorithm of feeder fault location for single-phase-grounded fault with spur feeders is proposed through transform analytical method of equivalent length straight feeder with the same parameters. According to operation experience and the principle of nonce feeder multi-phase-fault current protection (MPFC-protection), a new principle of feeder MPFC-protection based on wavelet measure concept is creatively proposed in order to improve protective action speed and protection range. It takes full advantage of facilitating action of TCC for expanding protection region and improving sensitivity, with modifying protection performance and having fine practical values. Based on existing problems of studies about multi-phase fault location in distribution systems, multilevel distributed neural network (NN) locating model with the basis of feature band measure vector is creatively proposed: first level is NN model for solving fault resistance value; the second is distributed locating NN models with modifying faulty angle,according to partition section of fault resistance value and further according to partition section of faulty instant angle to same fault resistance section. A lot of simulations verify that it is validity and correctness, higher faulty distance measuring precision and have high theoretic significance and practical values.