Adaptive Resonance-Tuned Vibration Energy Harvester with a Variable-Inertia Architecture

This is the authors' abstract. We don't add key points for this paper.

Conventional vibration energy harvesters exhibit a pronounced reduction in output when the excitation frequency deviates from the structural natural frequency. To address this limitation, this study develops a variable-inertia piezoelectric–electromagnetic hybrid harvester in which a sliding tuning mass modifies the effective inertia and, consequently, the natural frequency of the cantilever system. A coupled electromechanical model and a finite-element modal model are established. An output-voltage-feedback bidirectional hill-climbing algorithm is implemented on an STM32 controller to identify a near-optimal mass position without directly estimating the excitation frequency. The first natural frequency is predicted to be 14.666 Hz when the tuning mass is located at the midpoint of the effective cantilever length. At this frequency, the output power first increases and then decreases with load resistance; at 700 kΩ, the root-mean-square (RMS) load voltage and mean load power reach 6.668 V and 63.520 μW, respectively. During a 10–20 Hz sweep with a fixed 700 kΩ load and adaptive tuning enabled, the RMS load voltage remains between 5.4537 and 6.668 V, while the mean load power ranges from 42.490 to 63.521 μW. Within 14–17 Hz, the output power remains above 93.838% of its peak value. These results provide a system-level proof of concept for the integrated variable-inertia architecture and output-feedback tuning strategy within the tested first-mode frequency range.