Abstract

Iteration is a major cause contributing to an increase in project duration and cost. By adopting the traditional models of design structure matrix and reward Markov chain, this paper proposes a hybrid algorithm for solving a task sequencing problem that aims for minimising the project duration. The proposed algorithm combines hierarchical clustering and genetic algorithms. This algorithmic strategy is intended to utilise the circuit concept to reduce the solution search space for GA. Our algorithm was compared to five other algorithms. Through numerical experiments, the proposed algorithm can solve large problems (number of tasks = 200), yield the same quality of solution results with shorter computational time, and deliver stable algorithmic performance.

摘要

迭代是导致项目持续时间和成本增加的主要原因。本文采用设计结构矩阵和奖励马尔可夫链的传统模型，提出了一种混合算法来解决以最小化项目工期为目标的任务排序问题。所提出的算法结合了层次聚类和遗传算法。该算法策略旨在利用电路概念来减少 GA 的解决方案搜索空间。我们的算法与其他五种算法进行了比较。通过数值实验，该算法可以解决大型问题（任务数= 200），以更短的计算时间产生相同质量的求解结果，并提供稳定的算法性能。