Improved piezoelectric energy harvester design using aluminum nitride for improved voltage and power output

This research focuses on improving the performance of piezoelectric energy harvesters (PEHs), which convert ambient kinetic energy into electricity. One of the primary challenges with piezoelectric harvesters is their high resonant frequencies, which often do not align with the lower natural frequencies of ambient vibrations, limiting their efficiency. The goal of this research is to propose a new technique to optimize the design of PEHs, enhancing voltage output and power conversion efficiency. The proposed method combines an Arithmetic Optimization Algorithm to optimize the harvester’s dimensions with a Dual Temporal Gated Multi-Graph Convolution Network (DTGMGCN) to forecast resonant frequency and harvested voltage. The principal objective is to reduce resonant frequency errors and enhance energy conversion efficiency. The results, implemented on a MATLAB platform, demonstrate that the proposed method outperforms the existing techniques, such as robust chaotic Harris Hawk optimization, K-Nearest Neighbor Algorithm, and Heaviside Penalization of Discrete Material Optimization. The existing techniques show errors of 0.04%, 0.06%, and 0.08%, while the proposed method achieves an error of only 0.02%. Additionally, in terms of efficiency, the proposed method reaches 98%, significantly higher than the 65%, 78%, and 85% achieved by the existing techniques. These findings indicate the efficiency of the proposed approach in improving the design and performance of piezoelectric energy harvesters, offering a promising solution for more efficient energy harvesting systems.