<p>The early detection of rolling bearing faults is a critical challenge due to the inherently weak transient impulses masked by environmental noise and periodic interferences in vibration signals. To address this limitation, this study proposes a novel optimized jump plus mode decomposition (OJPMD) framework incorporating Harris hawk optimization (HHO) for adaptive parameter selection, thereby significantly enhancing the precision of incipient fault diagnosis. In contrast to conventional decomposition methods reliant on manual parameter tuning—a process prone to suboptimal performance—the proposed OJPMD uses HHO to automatically identify the optimal parameters of JPMD by minimizing the average envelope entropy (MAEE), thereby reducing mode mixing and noise sensitivity. Subsequent to decomposition, the envelope analysis is applied to the decomposed intrinsic mode functions (IMFs) to highlight weak fault-induced transients and to extract characteristic frequencies. Validation using two real-world datasets shows that the OJPMD method outperforms traditional JPMD and other advanced techniques in noise suppression and the extraction of early fault features. Case studies reveal that the proposed approach achieves higher accuracy in fault frequency identification under high noise levels, highlighting its robustness and practicality for industrial predictive maintenance. This work advances fault diagnosis methodologies by bridging adaptive optimization with signal decomposition, offering a reliable solution for detecting incipient mechanical failures in noisy environments.</p>