差异演化算法及其在机械优化设计中的应用

Modified Differential Evolution for Constrained Optimization and Its Application

针对约束优化问题,提出了一种适于约束优化的自适应交叉率中心差异演化算法。在约束处理上采用只更新可行域内的点的方法,避免了传统的惩罚函数方法中对惩罚因子的设置,使算法的实现变得简单,同时在差异演化算法中引入群体中心点参与群体最优点的竞争,并且对交叉概率进行动态调整。仿真实验结果和工程计算实例表明,提出的算法具有较快的收敛速度和较好的稳定性、鲁棒性。

A center differential evolution algorithm with adaptive crossover factor for constrained optimization modified differential evolution algorithm （ACFCDE） was proposed to solve constraint optimization problems. The algorithm used three simple selection criteria based on feasibility to guide the search in the feasible region. The proposed algorithm did not adopt the penalty function method, in contrast to the penalty function method, the constraint-handing technique of this algorithm was very simple, it did not require additional parameters. In addition, the center point of the population was incorporated into the DE algorithm, which only participated in the competition of the best point with the other individuals of the population, did not in any differential evolutions. And the crossover factor of DE algorithm dynamically and linearly was modified. As these measures were adopted, the stability, robustness and global searching performance of DE algorithm have been improved greatly. Results of simulations and comparisons with the other algorithms based on four testing functions demonstrated the effectiveness, efficiency and robustness of the proposed ACFCDE.