The Improved Genetic Algorithm Based On Hybrid Coding Applied In Reactive Power Optimization Of Power System

With the rapid development of our electric industry, electric customers higher demand for the quality of electric energy since the21st century. Ensuring the modern power system security, stability and economic operation has been a serious and critical problem for the contemporary power workers. Reactive power optimization of power system is an important way to ensure power system security, stability, economic operation, which can reduce the active power loss of power system effectively and improve the voltage quality of power. Hence, research for reactive power optimization has great significance in theory and application.Reactive power optimization problem in power system is a mixed nonlinear optimization problem with many variables and constraints, the operating variables include continuous and discrete ones, so the optimization process becomes quite complicated. Traditional reactive power optimization algorithm relies on a precise mathematical model, continuous and derivable objective function, and cannot solve discrete variables accurately, which resulted in a larger error in solving reactive power optimization of power system that contains a large number of discrete variables impacting the accuracy of the calculation result. Artificial intelligence optimization algorithm does not require a precise mathematical model, and it is a better way to deal with on-linear and discrete problems. This paper briefly introduced two methods of encryption algorithm, analyzed the advantages and disadvantages of classical optimization methods and artificial intelligence algorithms, combined with the actual power system, and finally selected the genetic algorithm as a method of solving the power system reactive power optimization.Considering the characteristics of the reactive power optimization of power system, the paper optimized the structural dimension of the shearing system active power loss of power system. In this paper, the minimum loss of the active power was regarded as the objective function, penalty functions were used as constraints of power system state variables, fast decoupled method was used in Power Flow Calculation with greater calculating speed and high accuracy. Besides, basic principle and some improving methods of Genetic Algorithms are expounded in this paper. Improving methods of simple Genetic Algorithms is proposed to overcome premature convergence and slow convergent speed in the later evolution process of simple genetic algorithm. Using mixed encoding of integer and real number to deal with discrete variables and continuous variables in reactive power optimization of power system; Using row re-operation that increases the diversity of the population in the initial population when it is generated; Chromosomes with high function value were adopted optima genetic algorithm to avoid losing in the operation; Gradually reducing the crossover rate and appropriately increasing mutation rate in the process of evolution, which reflects evolution process of species in nature; Secondary mutation operation were done once in a while, not only to avoid a lot of repeated computation but also increase the direction of the optimization, improving the ability of local search of the genetic algorithm.In order to testify the validity and feasibility of improved genetic algorithms in this paper, the simple genetic algorithm and the improved genetic algorithms are programmed by MATLAB language. The proposed algorithm is applied to the IEEE14-bus system to calculate reactive power. The results show that, compared with the simple genetic algorithm, Improved Genetic Algorithm used in reactive power optimization in this paper can effectively reduce the system active power loss, higher convergence precision and better global convergence, under the premise of ensuring the compliance voltage of power system.