Identifying Critical Category Combinations of Bridge Cables at Risk of Internal Resonances Using Data Mining Technology

Fan-shaped cable nets in cable-stayed bridges (CSBs) exhibit a dense spectrum of local frequencies, some of which inevitably lie close to global frequencies. This proximity increases the risk of internal resonances (IRs) in the system, posing an ongoing challenge due to their highly nonlinear nature of the dynamics and multimodal interactions. This study identified critical category combinations of cables at risk of in-plane vertical IRs using a data set comprising 2,437 real samples, aiming to propose IR-risk assessment strategies for CSBs. The investigation began by constructing an original data set using validated algebraic and finite-element solutions. The IR frequency condition was determined through indoor experiments and used as a risk indicator to identify high-risk data sets. Subsequently, an improved Apriori-based model incorporating deterioration probability was developed to analyze associations among the cables at risk. Three critical category combinations and their respective risk transmission paths were identified, revealing potential association rules between the design parameters of cables at risk and those of the entire bridge. In particular, cables anchored near pylons in the main span of mega-CSBs demonstrated the highest occurrence probability. Notably, if such cables also exhibit high-elastic-support effects, they require increased attention due to their potential IR risk. Finally, engineering application strategies were proposed, offering practical insights for improving IR risk assessment in bridge engineering.