摘要
该文提出了一种基于优化算法的冲突仲裁策略 .在约束网络支持的并行设计过程中 ,建立了冲突仲裁的数学模型 ,设计了反映冲突严重程度的指标 ,并针对该指标利用基因算法进行优化求解 .与以往的冲突解决方法不同 ,该方法利用设计参数区间的可扩展性 ,找到了制约设计问题的瓶颈 ,以最小的设计区间损失争取到最佳设计指标 ,从而达到有效解决冲突的目的 .与约束网络相配合 ,该方法能够在并行设计的初期就监控设计过程 ,提早解决冲突以避免后期出现大的返工 .
A conflict resolution strategy based on optimization is proposed. As opposed to the traditional fuzzy programming method, the method determines the conflict bottleneck and solves it at the least cost of design changes with the extendability of the design intervals. A conflict relaxation model is designed to meet the conflict resolution requirements, which extends the conflict degree to a continuous state from the on off condition of satisfy and conflict, then builds an objective describing the present conflict degree. A theorem stating that a more satisfactory result can be obtained by extending the corresponding design intervals is also proposed, which is the theoretical foundation for relaxing constraints. With constraint network, the quantitative effect of the downstream constraints can be considered before determining the design parameters. Furthermore, potential conflicts can be solved to avoid significant redesign. A bogie design example illustrates the usefulness of the method.
出处
《计算机学报》
EI
CSCD
北大核心
2002年第1期57-62,共6页
Chinese Journal of Computers
基金
国家自然科学基金 (6990 40 0 5 )资助
