Bai, Chin-YuChin-YuBaiLumentut, Mikail F.Mikail F.LumentutYI-CHUNG SHUTol, SerifeNouh, Mostafa A.Yang, JinkyuHuang, GuoliangLi, XiaopengChen, YangyangSugino, Christopher2025-07-072025-07-072025-03-17https://www.scopus.com/record/display.uri?eid=2-s2.0-105007439544&origin=resultslisthttps://scholars.lib.ntu.edu.tw/handle/123456789/730600This paper proposes an innovative approach to investigate the impact of electrode patterns on the energy harvesting performance of piezoelectric plates across various vibration modes. By integrating a conditional generative adversarial network (cGAN) with electromechanical finite element analysis (eFEA), realistic synthetic datasets are generated to evaluate the energy harvesting efficiency. The results demonstrate that cGAN-generated datasets significantly enhance the accuracy of power output predictions, with the proposed model achieving the average of Structural Similarity Index Measure (SSIM) values exceeding 0.9. The findings also highlight the critical influence of electrode configurations on modal frequency responses. Additionally, preliminary results suggest that this approach could be extended to segmented piezoelectric layer patterns on the plate.conditional Generative Adversarial Network (cGAN)deep learningelectrode patternselectromechanical finite element analysis (eFEA)piezoelectric energy harvestingplate vibration modesconference paper10.1117/12.3051289