A Study Of Distribution Network Reconfiguration Base On The Improved Patheno-genetic Algorithm

Distribution system is an important part in power system which connects power production to consumer. Under such the situation that economy is developing rapidly and power supply is short day by day, the reconfiguration of distribution network can improve security and economy of power system and exert the potentiality of existing distribution network through distribution network reconfiguration, which has great economic and social benefits.The advantage and disadvantage of traditional distribution network reconfiguration methods as well as intelligent bionic algorithms are summarized; it discusses the home and foreign methods of distribution power calculation and especially highlighting the technique of the backward/forward iterative algorithm; the steps and factors of traditional genetic algorithm are described, and several improved genetic algorithms are presented, then the partheno-genetic algorithm are materially researched, which has the same fitness function, selecting method, terminating condition as the simple genetic algorithm, but is different from simple genetic algorithm in encoding method, genetic operation and application field.Being an objective function both to maximize the reliability and to minimize the feeder loss, constraint conditions are considered as actual situation and operation limits, and transforming the muti-objective problem to single-objective one by weighting; In view to the low-efficiency of traditional genetic algorithm and the complexity of all kinds of improved genetic algorithm, an improved partheno-genetic algorithm based distribution network reconfiguration is proposed. This algorithm reduces the length of chromosome through simplifying the structure of distribution network, then both encoding method and genetic operation base on loop is proposed. These strategies greatly reduce the production of infeasible solutions and enhance the efficiency of the algorithm. The rules of disposing on public switches is presented, which could make the computation more efficient. Test results indicate that the proposed algorithm is feasible and effective.