【摘要】 将标准遗传算法用于最优化问题时存在早熟收敛和后期收敛速度缓慢的现象。本文扼要分析了遗传算法的运行机制 ,提出一种基于小生境技术的改进遗传算法 ,应用种群中最佳个体的马尔可夫链模型从理论上论证了该技术维持种群多样度的有效性。对复杂函数的遗传优化仿真实验数据表明 ,改进的遗传算法不但具有良好的全局收敛可靠性 ,而且具有快的收敛速度。更多还原

【Abstract】 Canonical genetic algorithms have the defects of pre maturity and stagnation when applied in optimizing problems. This paper analyzed briefly the work mechanism of the canonical genetic algorithms and the causes of the stagnation, a class of variation based on niche technology employing crossover of individuals with similar fitness and (2+2) selection was proposed, which can improve the searching stability and increase the diversity of individuals in populations. By means of Markov chain model of the best individuals in populations, it was theoretically proved that the improved genetic algorithm can reliably maintain the gene diversity of populations in the evolution process. The emulation experiment data of optimizing complicated mathematical functions showed that the improved genetic algorithm not only have higher global convergence reliability, but also higher converging velocity. Applying the proposed algorithms to optimize the reliability design of a jack clutch, the results excel obviously that of constrained random direction method.更多还原