Power-Interruption Fault Diagnosis Algorithm For Power Distribution Network Based On Chain Table

Traditional fault diagnosis systems apply matrix to describe their topology structure.Because the matrix can’t apply the power distribution network’s many important characteristics, such as tree structure, few loop, the related algorithms based on matrix have the disadvantages of bad reconstruction ability, low efficiency and high space complexity. Therefore, with the chain table running through the full text, this dissertation proposed a set of comprehensive algorithm system for the power-interrupt fault diagnosis.This system includes the fault location for measured area, the fault location for unmeasured area fault location, and the short circuit calculation.In the part of the fault location for measured area:The topology input file can only supply the correlative relationships of every node.So this dissertation will acquire the father or son relationship and the level of each node form topology input file by depth search. For improving the efficiency and the application scope of the fault location algorithm, this dissertation will proposes the Chain Table Algorithm: Based on the chain table model,branch is taken for the unit to be used for the first location and searching the fault branch at the first. Then the data inconsistency between the head node and the end node could be used for the second location according to the link relation of the nodes in the fault branch; Local reconstruction is used to treat the nodes without data. The distortion information reported by terminal monitoring node might lead to wrong conclusion. So The data correctness of each node is judged by the first location or the data relationship of the nodes in each furcation,on which some fault-tolerant mechanisms that could deal with one or more points fault and their correct rates will be discussed in this dissertation.In the part of the fault location for unmeasured area:Mostly use the information of trouble calls from different position and the rough set theory for fault location. This dissertation will analyze the disadvantages of the method based on the rough set theory and propose a new algorithm based on Bayes formula:In the single point fault environment, the subtree for each node is searched by the construct feature of dendriform network and then the user area set and the user area complementary set of each node is acquired; the cost function value of each node is calculated by each complaint information of the corresponding nodes in the user area set or the user area complementary set of this node, and the ambient area of the node that has the largest cost function value is regarded as the result. In the multiple-fault environment, each candidate fault group is composed by one or more component area that may by broken, and the user area set and the user area complementary set of this group can be aquired by calculating the intersection set of all the members’user area sets and the union set of all the members’user area complementary sets, then its cost function value can be caculated easily combining with the current complaint information sequence. The group with the largest cost function value and less members is regarded as the location result.In the party of short circuit calculation:Considerring the power distribution network’s characteristics such as large scale, dominate dendriform structure while few loop, lower voltage level, reverse charging current from synthetic load, and so on, the factors including accuration, efficiency, huge scale data procession, resistance, reverse charging load should be considerred in the algorithm design. So a new algorithm for short circuit calculation is proposed in the dissertation:Based on the theory of chain structure, the unique path between two arbitrary nodes in network and Thevenin’s Theorem. It acquires the value of short-circuit current by means of combinate powers pairwise.Then extend the design to complex network with loop(s):translated complex grid with several loops to some dendriform subnets with substitution theorem, and then acquired voltage matrix equation by Thevenin’s equivalent on power paths, calculated the current of broken node sequentially.