Global Performance of Tension Leg Floating Offshore Wind Turbine Based on Quasi-Dynamic Method and LSTM Method

Abstract The global motion response of the floating offshore wind turbine is critical to the tension of the tendons under the complicated environment. Several factors should be taken into account including the tidal effect, tendon pretension, set-down effect, the 1s and 2nd order wave loads, wind and current loads etc. A quasi-dynamic method is proposed to deal with the motion of the structure supporting the wind turbine and tension response of the tendons. Hydrodynamic coefficients including the added mass, radiation damping coefficients, the 1st order wave loads RAOs, 2nd order difference-frequency, sum-frequency wave loads are calculated based on HydroStar. An analytical expression between the vertical motion and horizontal offset is proposed and employed when estimating the tendon tensions. WF motion response is reconstructed based on the RAOs. The global performance of the wind turbine based on the quasi-dynamic method is compared with the results obtained based on Long-short Term Memory (LSTM) neural network method. It is verified that the satisfied results can be obtained based on both methods when compared with the traditional method, which means that it is feasible to find the surrogate model that is more efficient, especially considering the large amount of cases in the real engineering applications.