Optimal Reactive Power Planning Of Radical Distribution System On Improved Genetic Algorithm

The reactive power optimization of a radial distribution system can efficiently minimize the real power losses and improve voltage level of it. so that the yearly running charge will be reduced and the quality of power energy can be increased. There are two directions to research in the reactive power optimization. One is that the mathematical model used in reactive power planning is in line with realistic situation, namely objective function and its constraints are in accordance with realistic situation. The other is that the solving methods are modified to deal with the problem of local optimum and computational burden.Power flow method of a distribution system is the base and tool of reactive power optimization, whose convergence and computation speed is very important to the efficiency of optimization. This paper programs two different typical methods, and makes a comparison and analyses of them through 29-node, 33-node and 69-node feeder.In this paper, we present the objective function considering energy loss and investment of reactive compensation equipment at diverse load levels, Penalty function is considered to deal with variables violating the constraints. The mathematical model is established which takes all practical constraints and reactive power regulation measures into considered. The candidate locations for capacitor are situated in the end of lOKv line, a genetic algorithm with sensitivity analysis is designed to limit the number of candidate locations for capacitor. At the same time, in order to decrease the searching time of GA , self-adaptive crossover rate and mutation rate are introduced, in order to reducing the search space for the optimization algorithm we consider a big population size in the beginning several generations in order to cover more space, a small population size is selected later in order to increase the convergence speed..Based on the proposed mathematical model and algorithm, a computer program is developed. 33-node , 69-node feeder and Yinchuan radial distribution system are studied. All the results given in the paper show that the genetic algorithm with sensitivity analysis is stable and fast and suitable for practical application medium-scale power system.