Recognition Method of Turbine Pollutant Adhesion in Tidal Energy Generation Systems Based on Deep Learning

作     者：Mo, Caixia Zhu, Wanqiang Lu, Bingquan Zu, Shuai Zhang, Fuli Chen, Jianmei Zhang, Xiao Wu, Baigong Zhang, Xueming 

作者机构：School of Physics Northeast Normal University 5268 Renmin Street Jilin Province Changchun China Jilin Provincial Key Laboratory of Advanced Energy Development and Application Innovation 5268 Renmin Street Jilin Province Changchun China College of Mechanical and Electrical Engineering Central South University Changsha China Changchun University of Science and Technology China Jiangsu University of Science and Technology China 

出 版 物：《SSRN》 

年 卷 期：2023年

核心收录：

主　　题：Image enhancement 

摘      要：In light of the extended exposure of tidal energy generation systems to seawater, which can lead to marine biofouling and subsequent reductions in turbine efficiency, this paper presents a deep learning-based method for the identification of pollutant adhesion. This method aims to rapidly assess the adhesion status of turbines in tidal energy generation systems. A dataset of adhesion images from tidal energy generation system turbines under varying levels of pollution was obtained through underwater experiments with artificial fouling. Three different deep learning algorithms were employed to investigate the enhancement of underwater biofouling image quality. Image segmentation algorithms were used to extract and identify information related to the location and area of biofouling. The results demonstrate that the proposed deep learning-based pollutant adhesion identification method effectively recognizes the adhesion status on turbine blades, improving the accuracy of pollutant identification. This approach provides an efficient and accurate means of pollutant detection and management for the operation and maintenance of tidal energy generation systems, ultimately reducing operational costs. © 2023, The Authors. All rights reserved.