Based On Dynamic Programming And Genetic Algorithm Optimal Capacitor Placement For The Suqian Distribution Network

With the rapid development of economy and change of manage method of power system operation and programming of distribution power system are facing new problems. Reactive power compensation is an important approach to improve the voltages of distribution system and reduce energy losses. In our country capacitor compensation is a main method in the 10kv level distribution system. This paper presented a new method at the optimal placing of capacitors in the distribution system based upon dynamic programming and genetic algorithm (GA) to achieve the best economic behavior with eligible voltage.The sum of power losses and investments in the new adding VAR equipment which has been divided in each year is used as objective function. At the same time ineligible voltage is considered by a punishing term. We resolve the complex mathematical model using dynamic programming and genetic algorithm. Equality and inequality constraints make the model more comply with the actual situation and avoid the limit of single load operation condition. This algorithm is more rational and comprehensive in the evaluation of power system security and economy.Dynamic programming is a method for excessive phases optimal decision---making procedure. The reactive power compensation problem is disassembled to many subproblems and resolved one by one. Every phase should make its decision and this decision will be used as initial values for next decision. When decision of every phase is based on optimization the best goal is got in the final. In this paper sensitivity ?ΔPL ?Qiis used to measure the effect of unit reactive power compensation on power losses. This algorithm has approving speed. But because some predigestion is introduced in sensitivity calculation and analyse is based on local nodes the result is not very precise.Compared with the classical analytical methods genetic algorithm has less confinements about the problems. The only requirement is resoluble. It has no requirement for differentiable or continuity. The objective function also needn't to be convex function. GA starts from multi-points in the best goal seeking procedure. It can adjust parameters continuously and has stronger ability to reach the optimal goal in whole space. In the calculation we considered the trait of distribution network of the city of Su Cheng. For example the distribution system often has radiate structure and has less long transmission lines and X/R is small. The forward-backward sweeping algorithm is used to calculate the power flow of distribution network. It has the excellence of calculating quickly and engrossing less memory while avoided the dependence for system traits as other methods. The new algorithm gained from the combination of dynamic programming and GA has satisfactory speed and enough precise.The concept of equivalent maximum loss hours is used to coordinate multiple load levels in this paper. The costs of new installed compensation equipments and the effect on power losses are considered synthetically, avoided the limit of considering only one load level and got more precise results. Simulation cases show that the programming algorithm presented in this paper works well in getting the target, measuring the economic benefit. It has better calculation efficiency and stability with less confinement on the problems.