An Improved Real-Coded Genetic Algorithm and Its Application

value firstly. Computation efficiency will decrease. Abstract⎯Real-coded genetic algorithm (RGA) usually meets the demand of consecutive space problem. However, compared with simple genetic algorithm (SGA), RGA also has the inherent disadvantages such as prematurity and slow convergence when the solution is close to the optimum solution. This paper presents an improved real-coded genetic algorithm to increase the computation Based on these points, real-coded genetic algorithm appears, which can avoid some disadvantages of binarycoded genetic algorithm, but its inherent defects in algorithm also exist. In order to improve the performance of GA, a lot of research has been conducted. Ortiz-Boyer et al. proposed a way to improve crossover operation of realcoded GA and was successfully applied to solve artificial problems. Chang efficiency and avoid prematurity, especially in the optimization of multi-modal function. In this method, mutation operation and crossover operation are improved. Examples are given to demonstrate its computation efficiency and robustness. [12] contrived the multi-parents crossover to estimate the parameters of nonlinear process systems. Javadi et al. combined GA with neural network to deal with high dimension problems; meanwhile the computation efficiency was also improved. Index Terms⎯Adaptive mutation, arithmetic crossover, elitist strategy, genetic algorithm. In this paper, both mutation operation and crossover operation are improved to increase convergence speed and avoid prematurity. The organization of this paper is as follows. In section 2, the genetic algorithm steps are spread out. In section 3, we set forth the improvement on mutation operation and crossover operation. In section 4, two examples are given to testify the feasibility of IRGA. Conclusions are given in section 5.