Modified Kinetic-molecular Theory Optimization Algorithm And Its Application Research In Reactive Power Optimization

The optimization problem always is a hot research focus in scientific research and engineering applications.It is important to seek an optimization algorithm,which is simple,efficient and rapid,to solve the problem in the project.Intelligent optimization algorithm,due to it has many advantages,such as simple,good optimization effect and fast convergence speed,has attracted widespread attention of scholars.The distribution of reactive power in power system is unreasonable,not only affects voltage quality,but also makes power grid unstable,can't guarantee safe operation of power system.Reactive power optimization is an important method,which can increase the security of power system and decrease the network loss.Therefore,it is great significance to study this problem.However,reactive power optimization has many characteristics,such as multi constraint,nonlinear,multi variable,solving this problem using traditional mathematical method is unsatisfactory,so,it need intelligent optimization algorithm to solve reactive power optimization problem.In this paper,through the comprehensive analysis of the characteristics of many intelligence optimization algorithms,kinetic-molecular theory optimization algorithm is selected as the research object,then,analyzes its advantages and disadvantages,two improvement strategies has been proposed,finally,two modified algorithms are applied to reactive power optimization.Firstly,the correlation principle and operating mechanism of the kineticmolecular theory optimization algorithm are intensive studied,including the mechanism of attraction,repulsion and fluctuation among individuals,and the movement principle of best individual guided other individual in population.Two modified algorithms have been proposed to make up the defect of kinetic-molecular theory optimization algorithm,such as the lack of local searching mechanism,the wrong guide may occur,the solution accuracy needs to be improved,that are kinetic-molecular theory optimization algorithm based on crystallizing process and kinetic-molecular theory optimization algorithm based on teaching-learning process.A separation operator has been designed in C-KMTOA by simulating the crystallizing process and the population has been divided into three subgroups,the best individuals,the excellent individuals and the worst individuals,in addition,the worst individuals move toward the near of the excellent individuals,the excellent individuals move toward the near of the best individuals by the guiding operation,so that the search range has narrowed down the near of the optimal solution quickly.The teaching-learning process has been integrated in TL-KMTOA,in the process of optimization,the best individual can guide the individuals of subgroup move toward the optimal solution,individuals of subgroup can learn from each other to improve their fitness values,at the same time,the individuals which fitness is lower can improve the fitness values through two times learning of the best individual,this step is called feedback.Then,in order to verify the effectiveness of C-KMTOA and TL-KMTOA,two modified algorithms are simulated on the classical test functions.Experimental results show that two modified algorithms have shown some superiority in terms of convergence speed and accuracy.Then,two modified kinetic-molecular theory optimization algorithms are applied to solve the reactive power optimization.The solving method is analyzed,detailed calculation steps are given,and do the simulation experiments of reactive power optimization on the IEEE-14?IEEE-30 and IEEE-118 standard test system,the experimental results show that TL-KMTOA,TL-KMTOA algorithm has achieved good results for the optimization of the objective function,reducing the network loss,improving power quality,the effect of optimal is better than KMTOA,PSO,DE,TLBO algorithm.C-KMTOA and TL-KMTOA have been proposed to make up defect of KMTOA algorithm,both of that can improve the accuracy of solution,avoid algorithm stuck into local optimal.Finally,the modified algorithms are applied to reactive power optimization,and the feasibility and practicability of C-KMTOA and TL-KMTOA are verified by three examples.