High Voltage Distribution Network Reactive Power Optimization Run Research

Distribution network includes high voltage distribution network and low voltage distributionnetwork. Their network structure usually consists of loop network and radiation tree network. It isknown that the high voltage distribution network is loop network or the so-called closed distributionnetwork. The reactive power optimization and reactive compensation of power system aresignificant for the safe and economic operation of the power network. In the reactive poweroptimization of the distribution network, its principle means is to regulate the on-load tap changingtransformer and capacitors in parallel. In this thesis, we study the power flow calculation of highvoltage distribution network and the optimal reactive power compensation with the distributionnetwork of County as background.Firstly, we proposed the improvement program for international authoritative open sourcepackage Matpower. High pressure closed power flow calculation is very complicated and verydifficult to develop on their own, and using a commercial software needs a high investment and itsinterface is very complicated. Therefore, we adopt the open source Matpower power flowcalculation model. Recall that the latest Matpower can only be applied to the double windingtransformer, while the reality is the high voltage distribution network contains many three-windingtransformers. Based on this observation, we first introduce an equivalent model of three-windingtransformers, which makes Matpower can deal with the distribution network containingthree-winding transformers. This equivalent method can be verified by loop power distributionsystem with7nodes.Secondly, the improved genetic algorithm is proposed to solve the mathematical model forreactive power optimization of high voltage distribution network. Since the simple geneticalgorithm is slow in convergence and can easily be premature in the actual reactive poweroptimization of high voltage distribution network, which leads to local optimum solution rather thanglobal optimal solution, we propose an idea of mean value distribution for the generation of theinitial population in genetic algorithm. We also use the hierarchical and crossed methods for coding,and set the optimal retention operator with variant crossover rate and mutation rate, thus geneticalgorithm can be used much better to the reactive power optimization of high voltage distributionnetwork in reality.Finally, in this thesis, by using the example of the distribution network of County, we test andverify the improved genetic algorithm and prove the high efficiency and practicability of ourgenetic algorithm.