摘要
以Ｃ语言头文件方式开发成功遗传算法工具箱ＮＰＵＧＡＴｏｏｌｂｏｘＶ１．０。该工具箱既包括了遗传算法的基本算子，又包含了最新改进算子，可适用于单种群、多种群遗传算法。工具箱在ＴｕｒｂｏＣ２．０及ＢｏｒｌａｎｄＣ＋＋３．１环境下调试成功并通过不同算例测试。
Genetic Algorithm (GA) technology is rapidly emerging in China. To promote its widespread use, we applied for and were granted Chinese government support for research and development of GA toolbox. We completed R&D on NPU (Northwestern Polytechnical University) GA Toolbox V1.0 using C language. To satisfy the basic needs of GA application, we included in our toolbox four capabilities: parameter coding, random generation of initial population, scaling of fitness values, and optimization of certain functions via GA operators. Three options of parameter coding were offered: binary coding, integer coding, and real value coding. We proposed generation of initial population with random number. The user was provided with the choice of either linear or nonlinear scaling of fitness value. In addition to the basic operators, such as reproduction, crossover, and mutation, the toolbox also provided the quite new operators, such as twopoint crossover, inversion, deletion, and insertion. We utilized the concept of generation gap to improve the performance of our GA toolbox. We included migration operator in our GA toolbox to prevent the appearance of premature phenomenon. NPU GA toolbox V1.0 was developed and tested on a PC486 with Turbo C2.0 and Borland C ++ 3.1. We selected an arbitrary piecewise function with many peaks as shown in Fig.1. The function was optimized with binary coding (Fig.2) and real value coding (Fig.3) methods respectively. As expected, the two results agreed quite well.
出处
《西北工业大学学报》
EI
CAS
CSCD
北大核心
1997年第3期355-359,共5页
Journal of Northwestern Polytechnical University
基金
国家教委优秀青年教师基金
