A Reactive Optimization Model Considering Load Uncertainty

Optimal adjustment of reactive power can play an important role in secure and economical operation of power system,so that the research of reactive power optimization has the great significance in theory and practical application.As the uncertainty of loads that could cause the solution of reactive power optimization can not be the optimal one, it is necessary to consider the burden of uncertainty on the impact of reactive power optimization. The main work of this paper is as follows:â‘ First of all, the causes of the uncertainty of loads are analyzed, including the reasons of load forecasting model and the reasons of actual operation conditions. According to the random of the causes, a piecewise analytical model that analyzes the normal distribution of loads is built. The analytical expression of probability on each section is formed using the representation of each section and its corresponding probability.â‘¡Improved genetic algorithms for solving the solution of reactive power optimization are developed. Details of the development, theory and implementing process of Genetic Algorithms (GA) are presented. The fitness function, selection, crossover operator and mutation operator of the GA are improved. The fitness function that using real number code and linear scale change is adopted; the selection operator that combines the championship and the best individual is used; the mutation strategy that includes phases is adopted. Examples show: the improved genetic algorithm can effectively escape from local optimal solution.â‘¢Considering the impact of the load uncertainty on reactive power optimization, two kinds of reactive power optimization model are presented, according to the different treatment of load uncertainty; two models are called first subsection after total model and total combined model. A piecewise model according the normal distribution is adopted in the load model. In first subsection after total model, the probability of a load is included in the optimal solution; but in total combined model, the probability of a load is included in the fitness function. Analysis of examples shows: if load forecasting has errors, two kinds of reactive power optimization model in this paper can take into account the uncertainty effectively and to find the better solution of the reactive power optimization model with a single load. Comparing the model and optimal results, we can get the conclusion that total combined model is slightly better than first subsection after total model.â‘£Considering the parameters in the reactive power optimization, including the load section, constraints and penalty factors of the objective function. For different values of these parameters, the model is calculated and analyzed to analyze the impact of these parameters on reactive power optimization. Examples results show: number of load section would have greater impact on the optimal results, so we should choose appropriate number of load section; and changes of the scope of voltage constraint will have a greater impact on the optimal results; the value of penalty factor could express the optimization intent of decision-makers, different values of penalty factor can result in the different optimal results.