Study On Conductor Missing Diagnosis Of Grounding Grid Based On Co-evolutionary Genetic Algorithm

Grounding grid is an important part of power plants and substations. So it is vital to maintain its integrity. However, the conductor loss of new ground grid is widespread, because of the concealment of grounding grid construction and the lack of construction supervision. With the development of electric power system, most area of substations have been hardening, making it even harder to excavate. Therefore, it makes great sense in academic research and engineering practice to find out an intelligent optimization algorithm, so as to quickly locate abort situation without a power outage and the excavation of a large grounding grid.Co-evolutionary genetic algorithm introduced co-evolutionary ideas on the basis of the traditional genetic algorithm. In this paper, the algorithm is applied to the research of fault diagnosis resulting from the conductor loss. The main achievements are shown below:① To reduce the field measurements workload, but the measurements of port voltage can contains more branch information, according to sensitivity matrix between the the grounding grid port voltage and accessible nodes of the grounding grid to the conductor resistance. The screening principle is based on agglomerative hierarchical clustering results. Discarded nodes without quantitative requirements, if the number of nodes is still more after screening, port voltage related to each of the remaining nodes and cluster partition again, represent port can be chosen in the small area. Attention should be paid to the situation that the number of the ports in a certain area is rather small. At this time the area can not be ignored, to prevent the blind spots when gathering the conductor information. We first set a 4×4 regular grounding grid as the test object, then voltages of the ports optimized by clustering analysis and the same number of the port voltages measured by adopting the fixed point scheme are imported into the co-evolution algorithm diagnostic procedures. It turns out that a higher diagnostic accuracy can be obtained by adopting the voltages of the ports optimized by clustering analysis.② The number of grounding branch conductor have a greater impact on genetic algorithms in the diagnostic performance of the grounding grid conductors missing, so coevolutionary genetic algorithm introducts of co-evolutionary idea to improve the traditional genetic algorithm, branch conductors subdivided for reducing the population dimension algorithm iteration, improve computational efficiency and accuracy. According to the grounding conductor of the branch network port voltage optimized overall influence, clustered by agglomerative hierarchical method, in order to ensure partition reasonable and effective, the clustering coefficient consistent with the actual situation should above 0.9.③ Conductor loss diagnosis MATLAB program of grounding grid is written based on coevolutionary genetic algorithm, optimization measurement port and grounding conductors with reasonable partition. Simulation experiments are carried out based on the middle grounding grid of Dianfa substation and the big grounding grid of Shangqiao substation to test diagnostic results.of the program. It turned out that the coevolutionary genetic algorithm is better than the traditional genetic algorithm. Because the feasible individual was easier to appear in the iterative process, the fitness function value decreased faster, and the diagnosed performance has nothing to do with the area where conductor loss occurred. And the population size in the partition can be adjusted by the number of the branch conductor in the area. The effect of the diagnosis will be better, when the population size is 10~15 times larger than that of the branch conductor.