环境变异免疫克隆算法解决有约束优化问题

Environment Mutation Immune Clone Algorithm for Constrained Optimizations

为了克服免疫克隆算法搜索效率低、无法直接对进化经验学习等缺点,设计了环境变异免疫克隆算法,在普通免疫克隆算法中引入环境变异算子,通过环境变量积累进化过程中的经验,使算法具备了一定的学习能力;重新设计了代价函数,采用一种新颖的罚函数排序形式来处理由于约束条件造成的解集空间不连续问题,进而提高了算法的搜索效率及稳定性。通过对13个常用有约束优化问题测试函数的仿真实验,表明了环境变异免疫克隆算法在有约束优化问题上具有很好的性能。

In order to overcome some problems of immune clone algorithm, such as low efficiency and direct learning from evolution isn＇t available and so on, the Environment Mutation Immune Clone Algorithm （EMICA） was designed. A new environment mutation operator was introduced to normal immune clone algorithm, so self-stadying ability was obtained by accumulating the experience of evolution process with environmental variables. The cost function of constrained optimization was redesigned and a novel ranking approach for the penalty function was used to deal with the discontinuity of solution space caused by constraints, which resulted in the improvement of the searching efficiency and stability. Simulation results of 13 mark test functions demonstrate the good comprehensive performance of EMICA in constrained optimizations.