通过调节种群适应度改善遗传编程的收敛特性

根据遗传算法理论分析了遗传编程中种群适应度对算法收敛特性的影响,为改善遗传编程的算法收敛特性,提出了通过调节种群适应度进行改善的具体方法。以求证旅行商问题为例,通过统计性实验数据验证了改进后的算法较采用同样局部优化的常规遗传算法具有更好的收敛速度和优化解,并对改进后算法的相关控制参数选择进行了实验分析。结论为改进算法使遗传编程运行期能获得更好的收剑性能,该方法与具体求解问题无关,适用于常规遗传编程。

Improving Convergence Property of GA Programming by Adjusting Fitness of Population

This paper analyzes how and what population's fitness affects the convergence property of GA (genetics algorithm) programming according to basic GA theory. The method of adjusting the fitness of population, used to improve the convergence property of GA, is put forward. Taking an example of traveling salesman problem (TSP) to validate above-mentioned idea, improving algorithm has better convergence property in comparison with canonical GA through the experimental statistic data. The relative parameters for improving algorithm are analyzed and concluded also. The conclusion is that improving algorithm is independent of concrete problems and suitable for canonical GA programming, improving algorithm is to the good to convergence property and solution of GA.