基于遗传算法和蒙特卡洛模拟的并行工程设计工序优化

Design process sequencing optimization of concurrent engineering projects based on genetic algorithm and Monte Carlo simulation

并行工程是对产品设计及其相关过程进行并行、一体化设计的一种系统化工作模式,正作为工期压缩技术应用于建筑工程设计中。设计结构矩阵是可用于并行工程设计信息流分析的工具,然而,目前很多基于设计结构矩阵的模型并未考虑到成本和工期的不确定性,且无法很好地描述实际的并行设计过程。为解决该问题,提出了由遗传算法和蒙特卡洛模拟构成的工序优化模型,使并行工程设计过程中工序得到最优化。其中,遗传算法引入模型用于确定时间和成本目标函数下最优的工作顺序;蒙特卡洛模拟则用于解决工作时间和成本随机的问题。还引入返工概率、返工百分比、搭接矩阵、学习曲线和不确定的工期和成本等描述实际设计过程的特征,给出主要影响总工期和成本的返工量以及总时间和成本计算的具体算法。并应用于工程案例,输出结果验证了模型的有效性,为并行工程优化设计工序提供了新的有效的工具。

Concurrent engineering （CE） is a systematic pattern where the processes of product development and related processes are concurrent and integrated. CE is becoming one of the schedule compression techniques in construction design. The design structure matrix （DSM） is a tool that helps in understanding and analyzing the information flow in CE design. However, many existing DSM-based models do not consider the uncertainty of cost and time, and failed to adequately describe the realistic design process. An optimization model including Genetic Algorithm （GA） and Monte Carlo simulation （MC） is presented to obtain an optimization sequence of design process for CE projects. GA is used to obtain an optimization sequence, and MC is incorporated into the framework to tackle project activities with stochastic time and cost. An algorithm is developed to calculate the rework affecting the total time and cost. Rework probability, rework impact, overlapping matrix, improvement curve and uncertainty time and cost are introduced to represent the characteristics of design process. A case study verified the validity of the proposed model. The framework provides a new effective tool for design process sequencing optimization in CE projects.