Dual-mode optimization strategy of piezoelectric nanogenerators for wind-induced vibration energy harvesting

The growing demand for sustainable energy has intensified interest in piezoelectric energy harvesting from vortex-induced vibrations, particularly using bluff body structures exploiting the Kármán vortex street effect. However, the limited energy available for conversion necessitates robust structural design and circuit management strategies to minimize sensitivity to parametric uncertainties and external variations. Here, we propose a dual-mode optimization strategy to enhance the performance of a wind-induced vibration piezoelectric nanogenerator (WIV-PENG) by synergistically optimizing both the bluff-body structure and circuit management. The approach integrates Bayesian optimization (BO) for structural parameter refinement, yielding optimal dimensions of cavity depth (23.347 mm), opening diameter (21.523 mm), bluff body height (78.582 mm), and cantilever beam length (130.346 mm), alongside CNN-LSTM algorithm-assisted high-performance circuit management using a closed-loop BUCK-BOOST converter. This adaptive circuit management predicts time-series voltage data 1 s ahead to proactively adjust proportional-integral (PI) parameters, enabling dynamic pulse-width modulation (PWM) duty-cycle modulation for stable output. The optimized WIV-PENG achieves a peak voltage of 22.85 V and a maximum volumetric power density of 7.19 W/m3. Additionally, a bidirectional LSTM (BiLSTM) network enables real-time wind-intensity classification across six wind-speed conditions, achieving 99.17% accuracy. This dual-mode optimization framework significantly enhances the efficiency of vortex-induced energy harvesting, establishing a foundation for advanced piezoelectric energy-harvesting applications.