Research On Reactive Power Optimization Of Power System Based On Improved Genetic Algorithm

The rapid expansion of China’s power grid scale and the increasing complexity bring more randomness and unpredictability to the system operation,and also bring more severe challenges to the problem of reactive power optimization.The reactive power optimization problem of power system is characterized by the coexistence of continuous and discrete variables,multiple constraints and nonlinearity,so the artificial intelligence algorithm can show better flexibility and robustness when searching for optimization,and can effectively solve the reactive power optimization problem.In this paper,the genetic algorithm(GA)is chosen as the research object and it is improved.At the same time,the existing reactive power resources and voltage regulation methods are combined to reduce the system network loss and improve the performance of the system.Firstly,the research background,significance and research status of reactive power optimization of power system are introduced,and various algorithms used for reactive power optimization are analyzed.Secondly,a mathematical model of reactive power optimization is established to minimize the active power loss.At the same time,the overstepping voltage of load nodes and the overstepping reactive power output of generators are added into the objective function in the form of penalty terms.Two commonly used power flow calculation methods are introduced,and the P-Q decomposition method is selected as the power flow calculation method in this paper.Finally,the characteristics and basic operations of simple GA are described.Considering the problems of easy convergence to local optimal solution and slow convergence speed when solving reactive power optimization problems,some improvements are made to the simple GA.The control variables are coded by integer real mixed number.In the aspect of generating initial population,the randomly generated individuals are selected roughly.A hybrid choice method of roulette and elite retention strategy is adopted.Two aspects,individual fitness value and evolution algebra,were considered in the dynamic adjustment of parameter crossover and mutation probability.The improved GA and simple GA are applied to the IEEE-30 standard node system and the actual node system in a certain area.The simulation of the IEEE-30 node system shows that the active power losses of the two algorithms are 7.05 MW and 6.83 MW,respectively.The improved GA has better loss reduction effect and faster convergence speed,which verifies the feasibility and superiority of the improved GA in this paper.After the reactive power optimization of the actual node system in a certain area,the active power losses obtained by the two algorithms are 17.18 MW and 16.86 MW,respectively.The loss reduction effect of the improved GA is better,and the optimization results of the improved GA are generally better in terms of voltage quality,which indicates that the improved GA in this paper has certain practicability.