Reactive Power Optimization Of Distribution Network With Distributed Generations Based On Improved NSGA-? Algorithm

With the increasingly serious problems of fossil energy shortage and environmental pollution,Distributed Generation(DG)has developed rapidly for its advantages of environmental protection and high efficiency.However,with the access of a large number of DG,distribution network transforms from passive network to active network connected by supplying the power and users.So,the DG will have a enormous impact the distribution network.In order to ensure the consideration of DG uncertainty in the research process,the output probability characteristics of DG are studied by using probability theory and method in this thesis.On this basis,the research on the correlation of DG is represented by wind power and solar power in the same region,and an improved multi-objective optimization algorithm with the Elitist Non-dominated Sorting Genetic(NSGA-II)is adopted to solve the reactive power optimization model of distribution network.The specific research work is as follows:DG have obvious randomness and volatility are represented by wind and solar,and there is also a significant correlation between wind and solar in the same region.These two points have a great impact on the safe and stable operation of distribution network,so the study of probability model of distributed generation is the basis of reactive power optimization of distribution network with DG.In this thesis,independent output probability models of wind power and solar power are established respectively by weighted Gauss mixture distribution and Beta distribution.Then,through the study of wind-solar correlation in the same region,the correlation between wind and solar is reflected by Copula function when the probability model of wind and solar power output is obtained,and the factors with affecting the safe and reliable operation of distribution network are taken into account by mathematical transformation to ensure the validity and accuracy of the following reactive power optimization research.Finally,the fitting effects of various probabilistic models are tested with the measured data of a certain area.On the basis of obtaining the joint probability model of DG output represented by wind and solar power,the solution method of multi-objective problem in reactive power optimization of distribution network is analyzed.In order to find the Pareto optimal solution set satisfying the distribution network topological constraints,an improved optimization algorithm in the crossover and mutation operation process based on NSGA-II algorithm is adopted.The improved algorithm effectively enhancesthe spatial search ability and convergence speed of the algorithm,guarantees the stability and diversity of the population,and makes the Pareto frontier distribution better.The simulation results verify the feasibility of the improved NSGA-II algorithm for solving Multi-Objective Optimization problems through the simulation analysis on the test function with combined with the evaluation index of optimization algorithm.In the research process of reactive power optimization,it is necessary to adopt appropriate methods to deal with the randomness of the load and the DG.To solve this problem,the thesis uses Markov-chain Monte-Carlo simulation method based on slice sampling algorithm to calculate probabilistic power flow,which not only effectively solves the stochastic problem of distributed generation and load but also improves the accuracy of probabilistic power flow calculation of distributed generation system.A multi-objective reactive power optimization model is established by the objective function of minimizing the active power loss and the total voltage deviation of nodes,and the improved NSGA-II is used to solve the optimization model.After the Pareto optimal solution set is obtained by the simulation of the improved IEEE33 node system,the same three groups of solutions are selected,and the data are analyzed respectively in the case of the independence and the correlation,which verifies the feasibility and validity of the proposed method in this thesis.