中文摘要：

为有效求解复杂约束优化问题，提出了一种基于Oracle的混合约束差分进化算法OBHSaDE。在OBHSaDE算法中，首先对Oracle罚方法进行了改进，并符合约束优化问题的求解要求。利用改进后的Oracle罚方法来快速找到问题的可行域，借助无约束优化算法SaDE能对可行域进行有效搜索，利用序列二次规划的超线性的收敛速度来减少评估次数和提高解的质量。仿真结果表明，改进算法不仅减少了评估次数、提高了解的质量，且具有很好的鲁棒性，还具有较少的用户参数，提高了算法的实用性。OBHSaDE是求解约束优化问题的一种具有竞争力的新方法。

英文摘要：

ABSTRACT：To solve complex constrained optimization problems effectively, an Oracle penalty method- based hy- brid constrained differential evolution algorithm, OBHSaDE, was proposed. In OBHSaDE, the original Oracle penal- ty function method was improved to satisfy the standards of constrained optimization problems. The improved Oracle method can find feasible areas quickly. The adaptive differential evolution algorithm SaDE can explore feasible areas effectively. And with the help of sequential quadratic programming, the improved solutions can be found with fewer number of function evaluations. Simulation experiments and compared results show that the proposed approach not on- ly can improve the quality of the solution and reduce the number of function evaluations, but also is robust. In addi- tion, due to this method has fewer user parameters, the practicality of it is enhanced. The proposed approach is a new competitive approach for constrained optimization problems.