Enhanced Genetic and Imperialist Competitive Based Algorithms for Reducing Design Feedbacks in the Design Structure Matrix
提出增强型帝国主义竞争算法、遗传算法及混合方法，用于优化设计结构矩阵中的活动排序，最小化反馈长度，提高复杂产品开发效率。
Nowadays, finding an appropriate interrelated activities sequence in designing complex engineering products, leading to efficient product development, is known as a major concern for managers. Recently, a variety of efforts aim to help the design managers tackle this issue, and one of the most useful management tools among them is the sequencing analysis method in an activity-based design structure matrix (DSM). The sequencing analysis method is the process of reordering the DSM rows and columns to minimize the detrimental effects of feedback loops in the design process. Therefore, in this article, an enhanced imperialist competitive algorithm (ICA), an enhanced genetic algorithm (GA), and a hybrid ICA–GA method are presented to find an activity sequence in DSM with a minimum total feedback length, which is a proper criterion for diminishing the detrimental effects of the feedback loops. To this end, we have improved the ICA and GA methods utilizing two techniques: applying the operators dynamically and tuning the main parameters adaptively. Subsequently, some experiments are designed to evaluate the proposed methods’ performance in terms of the cost value, computational time, and convergence rate on a remote sensing satellite DSM as a case study and eight other DSMs. The performance results demonstrate the superiority of ICA over two other methods. Eventually, an exhaustive comparison with the four well-known methods involving Antares, efficient-simple GA, insertion-based heuristic, and insertion-based simulated annealing illustrates the superiority of the ICA in large-scale problems, yielding the best-known solution in a reasonable computational time.