A method for module partition in undefined complex systems

Abstract

Module partition is a cornerstone of modular design. Currently, the majority of module partition methods are devised for established systems with preliminary designs in place. However, for systems in the conceptual design phase, where the interrelations among internal components are yet to be fully undefined, such as small modular reactors, there exists a dearth of mature partition methods. In this paper, a new partition method tailored for undefined systems is proposed. Firstly, the random network is utilized to describe the mathematical model of the undefined system. Based on this, the well-known community detection method, the Louvain algorithm, is introduced for module partition for the undefined systems. Furthermore, a new module partition scheme that integrates the Louvain algorithm with intelligent optimization algorithms is proposed to address the problem that the Louvain algorithm is prone to be trapped into local optima for complex-structure undefined systems. Several systems, including simple-structure systems and undefined complex-structure systems, are constructed for validation, and results demonstrate that the newly proposed module partition method significantly enhances modularity and achieves superior partition outcomes for complex systems compared to other existing methods.