A multi-objective parameter updating method for high-fidelity dynamic modeling of gear transmission systems

Dynamic models are widely used due to their strong interpretability and physical consistency, making them effective tools for analyzing and predicting system behavior. A high-fidelity dynamic model can accurately represent the system’s dynamic response. However, many existing approaches depend on idealized assumptions, which often cause notable discrepancies between simulated and measured vibration responses. Moreover, most current optimization methods emphasize time-domain features, while the role of frequency-domain characteristics remains insufficiently explored. Therefore, this paper presents a parameter updating approach for the dynamic model, aiming to improve model fidelity and predictive reliability. First, a dynamic model of the gear transmission system is established based on the finite element method. Nonlinear excitations such as geometric eccentricity and gear backlash are introduced to improve model accuracy. Sensitivity analysis is then conducted to identify the key parameters influencing the system’s dynamic response. Finally, a multi-objective optimization of these key parameters is performed using the NSGA-II. Experimental validation on a single-stage gear test rig demonstrates the effectiveness and practicality of the proposed approach in reducing response discrepancies and improving model accuracy.