Optimizing Design Structure Matrices Using Markov Chain Modeling and Community Detection

Abstract

This article proposes a new method to enhance the efficiency of the design process in systems engineering. Our approach involves utilizing a Markov chain based on the design structure matrix (DSM). By creating a transformation matrix using the DSM and converting it into a Markov chain, we enable faster convergence of process iterations and improved decoupling between modules or clusters. The Markov chain is plotted as a tree using a layered layout algorithm and partitioned into communities using the Louvain algorithm. We applied our method to two types of DSM data, i.e., component-based and parameter-based, and the results showed valuable insights into the design process. Our approach is a valuable tool for managing complex systems, especially as systems become increasingly complex and challenging for engineers to manage during the design process.