一种基于空间结构的数值优化演化算法

提出一种求解数值优化问题的演化算法－－基于空间结构的演化算法(Space GA），在这种算法中，作者将演化种群中的每个个体放在固定的位置上，杂交操作在其邻居上的几个点进行，因此不用选择遗传操作的父体，从而避免了确定选择压力的问题，同时空间结构保证了搜索的全局性，遗传操作保证了较优解在其空间中的扩展，从而达到了全局寻优的目的。文章还讨论了不同的空间结构算法的影响，此算法可以求角数学规划问题、约束函数优化问题，如果对实型变量采用取整的操作，算法还可以求解混合整数非性规划问题，数值试验的结果表明了算法在求解的速度，稳定性，质量等方面都优于一般的演化算法。

A New Evolutionary Algorithm for Number Optimization Based on Space Structure

We proposed a new evolutionary algorithm for number optimization based on space structure (Space GA ). In this algorithm, every individual of the population is put into a fixed location of the specific dimension space structure, crossover operator operates among its space neighbors, and the SpaceGAs avoid determining the selection pressure because the algorithm needn't select the father of genetic operators, at the same time the space structure ensure its global searching capacity, the genetic operator ensure the better solution diffusing in the space structure and shrink to the global optimization. The influence of different dimension space structures to the algorithm's performance is discussed. In the end the SpaceGAs are applied into multi|modal function optimization, mixed|integer programming and function optimization with constraint, the results show that the SpaceGAs are super to the traditional evolutionary algorithms in speed, stability and quality.