An Improved Adaptive Genetic Algorithm
Although the search process of GA may appear the global optimal solution, it can not guarantee that it is converged to the global optimal solution every time, but also the possibility of precocious defects occurs. For disadvantages of genetic algorithm, a
- PDF / 180,249 Bytes
- 7 Pages / 429.725 x 659.895 pts Page_size
- 91 Downloads / 248 Views
Abstract. Although the search process of GA may appear the global optimal solution, it can not guarantee that it is converged to the global optimal solution every time, but also the possibility of precocious defects occurs. For disadvantages of genetic algorithm, an improved adaptive GA is proposed with a real-coded, temporary memory set strategy, the improved cross-strategy and the improved mutation strategy. The results of demonstrate examples are proved that effectiveness of the improved GA is best. Keywords: GA, improved algorithm, temporary memory set.
1
Introduction
Genetic algorithm(GA) is based by Darwin’ evolution and Mendel’ heredity[1].Although simple genetic algorithm (SGA) is effective on many problems, that conclusion can be made by the theoretical proof is: It cannot guarantee that SGA is converged to the global optimal solution every time, because the global optimal solution of SGA search process can’t be retained, which SGA is directly related to the use of generation replacement technology. When parents are substitute by a perfect offspring, each generation of outstanding individuals may not produce offspring due to the probability characteristics of the search, which will result in slower searches and even than the most optimal solution[2].There may be shortcoming of SGA premature [3].
2
The Improved GA
Solving practical problems by GA, we hope that the optimal objective is converged in a wide range of space and the direction to the optimal solution as soon as possible, so the global optimal solution is found. In order to take into account both, an improved adaptive GA (IAGA) is proposed [4]. It is in the following areas: 2.1
Selection Strategy
Although the search process of SGA may appear the global optimal solution, it cannot guarantee that it is converged to the global optimal solution every time SGA [5]. In order to obtain the global optimal solution, the strategy of retaining the best individuals must be used. However, retaining the best individuals may lead to premature; the key H. Tan (Ed.): Knowledge Discovery and Data Mining, AISC 135, pp. 717–723. springerlink.com © Springer-Verlag Berlin Heidelberg 2012
718
H. Tang
reason is that the best retained individuals may be most or all of one super individual or very similar super individual. And when their fitness are very bigger than other individuals’ fitness, the descendants of the best will be mostly selected in accordance with the fitness of individual offspring. Next the algorithm can’t improve the fitness, so the algorithm can’t go on iteration, which only leads to converge to local optimal individual. In order to prevent premature, this reference, a set of temporary memory strategies to improve the GA performance is proposed by mechanism of immune memory [3]. When antigens invade the body again, high-affinity antibody can be produced more than that of the initial immunization, known as immune response again or immune memory. To remember enough antigens, limited resources (memory cells) cannot increase indefinitely. The str
Data Loading...
