Article Preview
TopIntroduction
Genetic algorithm is a probability search algorithm, enlightened by Darwin's evolution theory and Mendel’s genetics, simulates the creature’s genetic and evolutionary process. It has been drawn wide attention since Michigan University’s professor John Holland proposed the conception of Simple Genetic Algorithm(SGA) in 1970s. According to the struggle for existence principles,produced new generations and evolved to the optimal solution through a series of genetic manipulation such as selection, crossover and mutation by leveraging colony searching technology. Compared with other optimizing methods, genetic algorithm described the researching problem in the form of string, simply used fitness function for optimization, without the need of derivative or other auxiliary information, especially for dealing with complex and nonlinear problems, which traditional search methods were difficult to solve, can be widely used in combinatorial optimization, machine learning, adaptive control and image processing and other fields.
Although genetic algorithm has been successfully applied to many fields, but a lot of practice and research show that the simple genetic algorithm has poor search ability and premature convergence defects, especially in dealing with multimodal function, this problem has become more prominent. It can only find a few optimal solutions, and sometimes gets a local optimal solution, but we often want to optimize the algorithm to identify the optimal solution for all. The same problem solving many times may help us to find multiple solutions, but the solution results are random, we cannot guarantee to find all global optimal values. Some theoretical studies have proved that traditional simple genetic algorithm did not converge to the global optimum (Rudolph, 1994). To overcome this problem, many researchers improved the genetic algorithm from the encoding, genetic operators, population patterns, the producing way of next generations and other aspects (Schraudolph & Belew, 1992; Chen & Wang, 2009; Lis & Eiben, 1997; Kuo & Hwang, 1996; Zhang, Wang, Luo, & Cong, 2010; Song, Qu, & Han, 2003). Although the global convergence rate and optimizing efficiency has been improved in some degrees, it needs further research and development. Simple genetic algorithm simulates the genetic evolution process relatively simple, less considering the individual characteristics of age, gender, reproduction. Niche Genetic Algorithm (NGA) (Huang & Chen, 2004; Xie, 2005) has good performance when be used in solving the premature convergence defects. Therefore, this article drew lessons from the sexual reproduction, introduced niche technology in the genetic algorithm, proposed the Niche Genetic Algorithm Based on Sexual Reproduction (NGABAR).