Abstract: To address the complex operating conditions of deep groove ball bearings in practical industrial scenarios and the time-varying friction characteristics of their contact interfaces, a four-degree-of-freedom dynamic model for faulty bearings is developed by incorporating the time-varying friction coefficient and variable rotational speed conditions. First, by analyzing a frictionless ideal model, a constant-speed model, and the proposed model, the influences of the first two simplified models on the vibration response characteristics of faulty bearings are investigated. Second, to deal with the non-stationary vibration signals induced by variable speed, an order tracking algorithm is employed for resampling, and envelope order analysis is adopted to extract fault-related characteristic information. Vibration analyses are then conducted on bearings under different fault states. Finally, comparisons between simulation and experiment verify the accuracy and feasibility of the proposed dynamic model. The results demonstrate that the damping energy dissipation effect of the time-varying friction force tends to weaken and reduce vibration impacts. The combination of order tracking and envelope order analysis can eliminate the interference caused by rotational speed fluctuations and highlight the inherent fault order characteristics of bearings. In the envelope order spectrum, the fault characteristic orders corresponding to each fault and their multiples can be clearly distinguished. Compared with single faults, the vibration amplitudes of compound faults change significantly, and obvious differences in vibration amplitude exist among different fault combinations. This finding indicates that the dynamic response of compound faults is not a simple superposition of single-fault features, but a nonlinear consequence of the mutual coupling between individual fault excitations and time-varying friction characteristics. This study provides a more realistic modeling framework for faulty deep groove ball bearings under complex operating conditions, clarifies the combined effects of friction and speed variation on vibration responses, and offers an effective signal processing strategy for identifying fault characteristic orders from non-stationary vibration signals.