An Island Partitioning Method Based on Cloud Adaptive Genetic Algorithm

An island partitioning method based on cloud adaptive genetic algorithm is proposed. The traditional genetic algorithm is modified by making use of the trait of cloud theory, which shows both randomness and a tendency of stability. The traditional crossover operator is replaced by cloud crossover operator, aiming at improving global searching ability and avoiding falling into local minimums. The concept of gene unit and relative restrictions are proposed. Unnecessary searches are reduced notably by repairing infeasible solutions. The controllable loads and the borders of islands are adjusted after the primary island partitioning scheme obtained, through which the result can be further optimized around the extreme point. In this paper, the priority and controllability of loads are taken into consideration. This method also gives full play to tie switches to optimize the network. This approach has been tested by the modified American PG&E 69-bus system with DGs and the effectiveness has been proved.