Application of artificial fish swarm algorithm with the mechanism of local learning and reverse learning in the reconfiguration of distribution network containing DG
Abstract:The reconfiguration of distribution network is an effective means to optimize the structure of distribution network. Considering the economy of the system, a mathematical model with minimizing the network loss as objective function is established. An artificial fish swarm algorithm with efficient parallel optimization is adopted to solve the distribution network reconfiguration problem with DG. In order to overcome the problem "dimensionality disaster" caused by binary coding, three major network simplification principles are proposed to improve the computational efficiency of the algorithm. When the algorithm falls into the cycle of "premature convergence", the mechanism of local learning and reverse learning are introduced. Part of fish according to the differential results of the fish in optimal position dynamically adjusted the direction, and work in coordination with optimal population to strengthen the local search; another part of the fish along the worst position start to reverse learning, flee the local optimal area in time and effectively improve the diversity of population. In order to further accelerate the optimization efficiency of the algorithm, the adaptive adjustment is made in the the parameters of vision and step length. The accuracy and effectiveness of the proposed algorithm are verified by example analysis.