基于离散事件优化的流水线集成优化设计研究

发布时间：2018-06-18 23:54

本文选题：数学规划 + 数学启发式算法　；参考：《工业工程与管理》2017年05期

【摘要】：流水线集成优化问题包含三个子问题:工作站分配问题,工作量切分问题与缓存分配问题。同时解决上述三个子问题,研究满足目标生产率要求的成本最低的流水线设计。为了解决上述问题,基于离散事件优化方法,建立描述仿真优化的混合整数规划模型;由于精确模型的求解复杂度高,精确的混合整数规划模型被近似为线性规划模型;这一近似模型的精度不足,为了进一步完善模型,调整了近似模型的构造,并以此为基础开发了数学启发式算法。数值实验表明,与现有的仿真优化软件相比,该数学启发式算法可以在较短的时间内求得满足约束条件且更接近全局最优的解。还研究了集成优化问题中的工作量切分与缓存分配图样。
[Abstract]:Pipeline integration optimization problem consists of three sub-problems: workstation assignment problem, workload segmentation problem and cache allocation problem. At the same time, to solve the above three sub-problems, the pipeline design with the lowest cost to meet the target productivity requirements is studied. In order to solve the above problems, a hybrid integer programming model describing simulation optimization is established based on discrete event optimization method, and the exact mixed integer programming model is approximated as a linear programming model because of the high complexity of solving the exact model. In order to perfect the model, the structure of the approximate model is adjusted and the mathematical heuristic algorithm is developed. Numerical experiments show that compared with the existing simulation optimization software, the heuristic algorithm can obtain the solution which satisfies the constraint condition and is closer to the global optimal solution in a short time. Workload segmentation and cache allocation patterns in integration optimization problems are also studied.
【作者单位】：上海交通大学机械与动力工程学院;
【基金】：国家自然科学基金资助面上项目(61473188)
【分类号】：TB497

【相似文献】