As critical load-bearing components in rotating machinery, rolling bearings face significant risks of localized damage under time-varying operating conditions due to complex load fluctuations, posing severe threats to equipment operational safety. In response to the inherent limitations of traditional monitoring methods that rely on steady-state operating conditions, this study innovatively proposes an adaptive health monitoring framework based on time-series tracking. Firstly, the non-stationary time-domain vibration signals are converted into angular-domain steady-state signals by an equal-angle resampling algorithm, and then the fault eigen frequencies are precisely extracted by order spectral analysis. Secondly, the dynamic fault threshold model is established using the Weibull distribution, and the sliding window mechanism is introduced to achieve the dynamic update of the threshold. Validation is performed using the Ottawa dataset and the XJTU-SQ dataset. The results show that compared with the traditional method, the proposed method can sensitively capture the fault evolution mode and ensure the timeliness and accuracy of online monitoring. This research provides theoretical innovations and practical examples for intelligent maintenance of rotating machinery operating under complex conditions.