Applying a machine learning method for cumulative fatigue damage estimation of the IEA 15MW wind turbine with monopile support structures
Peer reviewed, Journal article
Published version
Date
2023Metadata
Show full item recordCollections
Original version
Ren, C. & Xing, Y. (2023) Applying a machine learning method for cumulative fatigue damage estimation of the IEA 15MW wind turbine with monopile support structures. IOP Conference Series: Materials Science and Engineering, 1294, 012014 10.1088/1757-899X/1294/1/012014Abstract
Offshore support structures are critical for offshore bottom-fixed wind turbines, as they bear nearly all the mass and loading of wind turbine systems. In addition, the support structures are generally subjected to a harsh environment and require a design life of more than 20 years. However, the design validation of the support structure normally needs thousands of simulations, especially considering the fatigue limit state. Each simulation is quite time-consuming. This makes the design optimization of wind turbine support structures lengthy. Therefore, an effective approach for estimating the fatigue damage of wind turbine support structures is essential. This work uses a machine learning method named the AK DA approach for cumulative fatigue damage of wind turbine support structures. An offshore site in the Atlantic Sea is studied, and the related joint probability distribution of wind-wave occurrences is adopted in this work. The IEA 15MW wind turbine with monopile support structure is investigated, and different wind-wave conditions are considered. The cumulative fatigue damage of the monopile support structure is estimated by the AK-DA approach. The numerical results showed that this machine learning approach can efficiently and accurately estimate the cumulative fatigue damage of the monopile support structure. The efficiency is increased more than 55 times with an error of around 1%. The AK-DA approach can highly enhance the design efficiency of offshore wind support structures.