Research And Application Of Genetic Algorithm In Reactive Power Optimization Of Power System

Reactive power optimization is a complex non-linear programming problem,which attracts much attention in scientific research because its objective functions and restraint conditions are non-linear and it comprises both discrete and continuous variables.So far there has not been a quick and practical method for reactive power optimization.The key problems of reactive power optimization lie in assuring the convergence of the algorithms,handling the nonlinear of objective functions and solving the discrete properties of the control variablcs.Based on the available references for reactive power optimization,this thesis establishes a multi-objective mathematical model with the smallest of active power loss and reactive power compensation capacity,which meets a variety of restraint conditions,and converts the multi-objective function to a single-objective function by weight value transformation for genetic algorithm optimization.Three power flow calculation methods are compared through analyzing and simulating.Based on the results of analysis and comparison,and taking into the characteristics of repeatedly power flow calculation in genetic algorithm,the fast decoupled method is utilized in the power flow algorithm,which is convenient to be programmed and improves the convergence rate.Considering the deficiencies of simple genetic algorithm and combining the characteristics of reactive power optimization,a lot of improvement is conducted for encoding,selection,crossover and mutation operations and convergence criterion.The integer and real number mixed coding method,adaptive cross and mutation probability;non-uniform mutation,the superiority protection strategy and catastrophe strategy are adopted.IEEE 14 and IEEE 30 are used as examples to simulation analysis.The results show that these improvements enhance the stability,computational efficiency,convergence rate and global optimization ability of the algorithm.