实数编码遗传算法能解性

Solving ability of genetic algorithms with real-code

通过分析一类标准的实数编码遗传算法算子作用机制及其交叉算子的局部搜索特性,提出了变异的方向性的概念,指出变异算子向各个方向变异并非均匀,而是以坐标轴方向分布最多.同时定义并分析了变量可分离函数优化问题.数值试验结果表明,实数编码遗传算法比较合适求解变量可分离的函数优化问题.

The action mechanism of operator of real-coded genetic algorithms（RCGA） was analyzed. The characteristics of local searching of cross operators was deduced. Directionality of mutation operator was defined. Direction of mutation was not uniform and its direction distribution was dominating at coordinate orientation. Special function optimiaztion problems with separable variables were defined and the peculiarity was analyzed. It is concluded that RCGA is appropriate for sloving function optimiaztion problems with separable variables relatively, which is also tested by numerical experiments.