A combined algorithm for reactive power optimization of distribution network

The simple genetic algorithm is constrained by its poor converging performance and readily leads to local optimization.So in this paper,a combined GA,which can retain the advantages of GA,simulated annealing(SA) and tabu search(TS),is introduced.At first,according to the value of individual fitness the adaptive crossover and adaptive mutation are performed,and heuristic border upon mutation can speed up the anagenesis.Thus the diversity of group is increased and the local optimum can be avoided by simulated annealing.Then the tabu search with a specifical probability is used to improve the local search capacity.To improve the computation speed and accuracy,decimal integer encoding and reserving the elitist are adopted.One real system is used to test the performance of the combined genetic algorithm.Test results show that the hybrid genetic algorithm possesses following advantages: good computation speed and global convergence and high calculation accuracy.