There are many efficient and effective constraint-handling mechanisms for constrained optimization problems. However, most of them evaluate all the individuals, including the worse individuals, which waste a lot of fitness evaluations. In this paper, halfspace partition mechanism based on constraint violation values is proposed. Since constraint violation information of individuals in current generation are already known, the positive side of tangent line of one point as positive halfspace is defined. A point is treated as potential point if it locates in the intersect region of two positive halfspaces. Hence, the region includes all these points has greater possibility to obtain smaller constraint violation. Only when the offspring locates in this area, the actual objective function value and constraint violation will be calculated. The estimated worse individuals will be omitted without calculating actual constraint violation and fitness function value. Four engineering optimization and a case study with the grinding optimization process are studied. The experimental results verify the effectiveness of the proposed mechanism.

对于约束优化问题，有许多有效的约束处理机制。但是，他们中的大多数人对所有个人进行评估，包括较差的个人，这浪费了很多适合度评估。本文提出了一种基于约束违例值的半空间划分机制。由于已知当前一代人的约束违规信息，因此将一点的切线的正侧定义为正半空间。如果点位于两个正半空间的相交区域中，则将其视为潜在点。因此，包括所有这些点的区域更有可能获得较小的约束违规。仅当后代位于该区域时，才会计算实际目标函数值和约束违反。在不计算实际约束违反和适应度函数值的情况下，将忽略估计的较差个人。研究了四个工程优化和一个磨削优化过程的案例研究。实验结果验证了该机制的有效性。