Electro-structural analysis and machine learning-based optimization of a coupled vortex-galloping laminated piezoelectric energy harvester

Abstract

The present work introduces a hybrid laminated composite energy harvester that couples vortex-induced vibration (VIV) and galloping to enhance energy conversion. A finite element–based electro-structural model is developed using Hamilton’s principle, incorporating fluid–structure interaction to predict electrical and dynamic responses. The effects of wake angle, coupling coefficient, and geometric ratios are systematically studied to identify their influence on voltage output. To accelerate optimization, an Elman-type recurrent neural network (RNN) is employed as a surrogate model within an improved particle swarm optimization (IPSO) framework. The RNN–IPSO approach efficiently identifies optimal configurations, achieving significant gains in output power. Experimental validation in a wind tunnel with a CFRP laminated beam and surface-bonded piezoelectric patches demonstrates a 50% improvement over the baseline model. Cross-ply orientation yields the highest energy output, while angle-ply arrangements increase RMS voltage by 10–15% due to enhanced flexibility. Overall, the study highlights a novel integration of VIV–galloping excitation and intelligent surrogate-based optimization, providing a strong foundation for next-generation self-powered sensing and renewable energy systems.