Mechanistic and data-driven modeling of diaphragm coupling misalignment dynamics

Current research on diaphragm couplings fully addressed the mechanisms of misalignment-induced excitations and transmission-speed fluctuations, mainly due to modeling and computational challenges. To address this gap, this paper investigates diaphragm-coupling misalignment dynamics and proposes a hybrid framework integrating physics-based modeling with data-driven methods to improve computational efficiency. A double-shaft system is modeled using finite elements, and representative analytical expressions for misalignment-induced excitations are derived from bolt-position kinematics and diaphragm deformation. A solution algorithm combining a dual-loop slow/fast time-scale scheme, model-order reduction, and neural-network–based force surrogates is developed to enhance efficiency. Simulations validate the model and algorithm in terms of accuracy and efficiency, quantify rigid–flexible interactions, and characterize transmission-speed fluctuations. Angular and parallel misalignment excitations are analyzed, revealing the emergence of subcritical resonances and integer-order harmonics. Experimental validation using a double-shaft–coupling test rig confirms the misalignment-induced speed fluctuations and rigid–flexible interactions, while vibration data show the subcritical resonances and additional integer-order harmonics. These results provide deep insights into diaphragm-coupling misalignment dynamics and establish a computationally efficient framework for rigid–flexible rotor system analysis.