Year: 2025
Type: article
Abstract: Due to the particularity of wind resources and different wind conditions in different locations, wind turbines accumulate fatigue damage at different speeds, and the existing methods fail to provide accurate and personalized fatigue damage evaluation. A fatigue evaluation method based on machine learning was established based on the connection bolts of wind turbine tower flanges. GH Bladed software was used to simulate and calculate the load time data of normal power generation conditions under wind conditions with different parameter combination. Then, fatigue damage of bolts was obtained using Schmidt-Neuper algorithm, wind condition parameters-fatigue damage data set was established, and fatigue damage was converted into equivalent fatigue stress (EFS). The mapping model of wind condition parameters and EFS is established based on various machine learning algorithms, and the corresponding fatigue damage can be obtained according to any combination of wind condition parameters. The results demonstrate that XGBoost algorithm achieves the highest accuracy in fatigue damage evaluation.
Authors: Shixiong Gao, Juncheng Liu, Xing Wang, Chentan Zhao, Feijiang Wu
Institution: North China Electric Power University
Language: English
Subfield: Mechanical Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W4409202764