Abstract: It’s inevitable to produce infeasible points when the author use particle swarm optimization to solve nonlinear programming problems，and it is very important to handle infeasible points for particle swarm opimization to get good optimization result.The authors propose an reasonable select method to deal with infeasible points based on object function value and violate degree，and apply an hybrid particle swarm optimization algorithm combined differential evolution to solve constrined optimization problems，the numerical experiment indicated that the algorithm is efficient.

Key words: particle swarm, differential evolution, premature

摘要： 用粒子群算法求解非线性规划问题时不可避免的会产生不可行点，处理好不可行点是粒子群算法取得良好优化结果的关键。依据粒子的目标函数值与违反约束的程度提出了一种处理不可行点的合理选择方案，并运用融合差分演化的混合粒子群算法求解约束优化问题，数值实验表明该算法的有效性。

关键词: 粒子群, 差分演化, 早熟