Reactive Power Optimization Based On Improved Genetic Algorithm In Low Voltage Distribution Network

The reactive power optimization can efficiently improve voltage quality of the power system and ensure the security and economic operation of the power system. Since long time ago we attached more importance to the reactive power optimization in high voltage electric power transmission. Many researches have studied on it and have obtained many achievements in practice. But the reactive power optimization of the distribution network didn't get enough recognition. Since our country has began carrying out the improvement the power network, many automatic systems and equipments of the distribution network have been introduced. But these equipments were only been used in improving the reliability of the power supply. Distribution system which faces to the consumers is an very important part of the power system. It's operation reliability and voltage quality are highly required. So it is very important that how to deploy the reactive power of the distribution system to reduce the power loss of the distribution system, increase the acceptable rate of the voltage and improve the economic operation.In this paper we firstly build a mathematic model of the distribution system, then analyze the power flow algorithm based on the structural feature of the distribution system, at last chose back-and-forth method as our power flow algorithm. After analyzing the traditional reactive power optimization and AI (artificial intelligence), we chose simple genetic algorithm as our base algorithm. But the simple genetic algorithm has many defects, so we put forward some methods to improve it.The improved methods are given as follows: optimize the initialization population to increase the fitness of the elders; the better chromosome reservation is used to increase the average fitness of the filial generation and speed up convergence; improve the fitness function which a dynamic retribution factor is used; dynamic crossover and mutation factor and inbreeding mutation method are employed to reduce the probability of partially converge; the inquiry of the repeat chromosome which avoid the repetitive power flow computation increases the efficiency of the calculation; do a local search of the convergence solution to found a better result to increase the accuracy of the result; a new convergence method which include a minimum reservation generation of the best result and a maximum evolution generation is used.To illustrate the development of the improved genetic algorithm, a sample distribution system has been chosen. The improved genetic algorithm program was used to calculate the example. It has been demonstrated that the improved method runs much faster than the original one and the result is more accurate. Therefore the new method is effective to solve the distribution reactive power optimization.