Identification of the influential parts in a complex mechanical product from a reliability perspective using complex network theory

Abstract

AbstractDue to the difference in the numbers and strengths of physical relationships among parts, complex mechanical products (CMPs) have community structure characteristics. There are often some influential parts in the community. Failures of these influential parts spread rapidly along the physical relationships between parts in the community, which seriously affects the reliability of a product. Therefore, identifying the influential parts in the community and adopting targeted measures can effectively improve the reliability and service life of a product. However, identifying the influential parts within each community in a collection of parts with complex relationships is very difficult. Thus, from the perspective of reliability, a method for identifying the influential parts of a CMP based on complex network theory is proposed and used to identify the influential parts in each community of products. First, weighted complex network (WCN) theory is employed to construct a CMP into a WCN model. Second, the complex network community detection method is employed to detect the community structure of the WCN model. Third, a modified LocalRank algorithm is employed to identify the influential nodes in each community, ie, the influential parts in each community of a CMP. Fourth, a modified susceptible‐infectious‐recovered (SIR) model is employed to evaluate the impacts of the influential parts. An analysis of a company's DC drill planetary gearbox shows that the proposed method is accurate and effective.