Research On Distribution Network Reconfiguration With Distributed Generation Based On Multi-Objective Genetic Algorithm

Distribution network is designed as a closed loop and operated as an open loop, including many closed section switches and open connective switches. When state of the switches are changed in some way, the structure of distribution network is changed therewith, and it can be operated with balanced load, lower loss, better voltage quality and service ability.With a lot of Distributed Generation(DG) accessing to distribution networks, it’s necessary to consider the influence of DG in load distribution, voltage and reliability. In addition, with the electricity market pushed vigorously, power distribution reliability gets to be more and more important as index for assessing electricity supply service quality. In this article, reconfiguration of distribution network with DG is proposed with the impact of DG taken into account, to optimize the economic and reliability index comprehensively. Where, instantaneous loss in steady state is taken as economic index, and Expected Energy not Supplied in fault conditions is referred as reliability index.In mathematics, distribution network reconfiguration is a NP-hard high-dimensional non-linear multi-objective optimization problem with a great solution space, and it’s difficult to get a good solution by using general optimization methods. Genetic Algorithm(GA) is an adaptive probabilistic search technology for problem of complex system optimization——its inherent parallel and probabilistic mechanism of search help to find the solutions quickly, and its evolution without derivation is very suitable for a non-linear optimization problems as distribution network reconfiguration. This article introduces the concept of Pareto optimal theory, proposing a DG-containing distribution network reconfiguration formula based on Multi-Objective Genetic Algorithm(MOGA)——in Pareto optimal operation, elitist strategy and fitness evaluation, impact of crowding distance is considered, maintaining the diversity of the population effectively; with the constraint of radial structure, a " randomly loop-avoid method" is proposed to generate the feasible initial population, besides, in crossover and mutation operations, gene block of loop are operated as a unit, and special operations are used for loops with public branches, to avoid the generation of infeasible solutions, improving the optimization efficiency. Examples of IEEE33nodes are calculated and analyzed, the network structures with excellent steady-state economic operational capability and stably power supply capacity are generated.