改进的实码加速遗传算法

对实码加速遗传算法(RAGA)8个步骤的局部参数进行修改,再对最后一次加速收缩后的区间用标准遗传算法(SGA)进行精细搜索。经过实例证明,改进后的算法计算机运行的次数减少,并且精度也得到提高。另外,对加速后的区间产生偏向最优点一侧的概率做了理论上的探讨,提出了把区间端点值重新赋给2个个体参加下一轮搜索。这样处理后避免舍去上次搜索的最优值,在一定程度上避免了某个变量的搜索区间在最优值一侧发生偏移。

Improved real coding based accelerating genetic algorithm

The real coding based accelerating genetic algorithm(RAGA) has the ability of dealing with complex optimization problems,and its capability of adjusting and compressing search domain is good,but the domain of the RAGA still reduces slowly. In this paper, part parameters in 8 steps of the RAGA are amended,and fine search is carried out with simple genetic algorithm(SGA) in the last reduced domain. Calculation examples prove that the run times of the computer decrease and the precision also acquires enhancement. Furthermore,the probability that the reduced domain leans to one side of the best pointer is discussed and it is brought forward that two variables can be endued with the both ends of the domain so that the two units can join in the next search. This disposal can avoid abandoning the best pointer of the last search and can also avoid that the search domain of some variable leans to one side of the best pointer.