Piezoelectric nanogenerators (PENG) have emerged as promising candidates for self-powered wearable electronics, yet their practical deployment is constrained by insufficient output performance, the need for post-polarization treatment, and limited charge extraction caused by the shielding effect at the electrode–polymer interface. Herein, we report an integrated flexible self-powered sensing device based on a ZnO/PVDF composite piezoelectric nanofiber membrane fabricated via electrospinning, coupled with a ZnO charge transport layer deposited by magnetron sputtering, forming a dual-layer ZnO architecture. Through finite element simulations and systematic experimental characterization, we investigate the influence of ZnO nanoparticle size and loading concentration on β-phase crystallization and piezoelectric output. The results demonstrate that ZnO nanoparticles with a diameter of 30 nm at a loading of 2 wt% yield the highest β-phase fraction of 84.93%, attributable to enhanced interfacial interaction, stronger molecular chain confinement, and more pronounced interfacial polarization effects associated with larger specific surface area. The ZnO charge transport layer, acting as an n-type semiconductor that forms a Schottky junction and built-in electric field with the nickel electrode while simultaneously improving interfacial contact, reducing charge traps, and providing stress buffering, further elevates the open-circuit voltage to 5.3 V and short-circuit current to 1320 nA under 9 Hz excitation, representing improvements of 17.8% and 68.8%, respectively, compared with the single-layer device. A maximum output power of 0.81 μW is achieved at a matched load resistance of 8.2 MΩ. The device successfully charges commercial capacitors and illuminates 9 LEDs, confirming its practical energy harvesting capability. Moreover, the device exhibits excellent flexibility, biocompatibility, and cycling stability over 6300 bending cycles, enabling accurate detection of wrist joint motions, gait patterns, and radial artery pulse signals with characteristic P, T, and D waves. This study delivers a feasible route to fabricate high-performance self-powered wearable sensors. It also possesses great prospects for mechanical energy collection and real-time physiological signal tracking.