An automatic genetic algorithm(GA)based method for optimal reconfiguration of distribution network is proposed. One of the problems in using GA to solve the network reconfiguration problem is that the individuals generated by crossover and mutation may not satisfy the radial constraint of the distribution topology. To overcome this problem

firstly

a fast automatic mesh check algorithm is developed to determine whether the generated individuals are feasible or not in the evolutionary process. Then

to replace the infeasible individuals

another automatic algorithm which can generate new feasible individuals is proposed. The algorithm can also be used to form the initial population automatically based on the topology searching which is usually done. Thirdly

a guided mutation algorithm is introduced to improve the convergence characteristic of GA. The algorithm not only ensures that the generated configuration is radial

but also guarantees that the mutated gene is the best one among all the feasible candidate genes. The proposed method improves the automatic degree of the GA reconfiguration problem and the converging performance of the GA can be improved dramatically. The results of the tests that are carried out on three distribution systems

which are IEEE 33-node system

PG&E 69-node system and 119-node system

demonstrate the effectiveness and efficiency of the proposed method.