Intelligent Prediction Of Fatigue Damage Of Fixed Offshore Wind Turbine Under Combined Action Of Wind And Wave

Offshore wind structures are subjected to wind and wave loads cyclically up to10~7 times during their 20-year design life,and the response of the structure under combined wind and wave action needs to be analyzed for fatigue damage assessment.Due to the complex stochastic nature of wind and wave fields,traditional methods for estimating fatigue damage in offshore wind turbines need to simulate the response of the structure under the combined action of multiple wind and wave fields,which is computationally expensive.To improve the efficiency of fatigue damage estimation for offshore wind turbines,researchers have proposed a combined load set and frequency domain analysis method.Combining load sets is done by combining load sets that produce similar outputs,thus reducing the computational effort.As the number of combinations increases,the resulting errors also increase,resulting in a more limited improvement in computational efficiency from the combined load set approach.The frequency domain analysis method assumes of linear superposition of responses,which makes it difficult to accurately assess the non-linear coupling effect of wind and wave loads when estimating fatigue damage in offshore wind turbines under the combined effects of wind and waves,and therefore its accuracy is difficult to guarantee.With the development of artificial intelligence techniques,its application to structural fatigue loading has shown better performance.The method uses fewer,intelligently selected samples to construct the relationship between load and response,and then efficiently simulates the structural response under the remaining load combinations,improving the efficiency of structural fatigue damage estimation while ensuring accuracy.Artificial intelligence algorithms such as Gaussian regression models,kriging models and artificial neural networks have shown good applicability in being used for structural response analysis under multiple environmental parameters and can be applied to fatigue calculations.The accuracy and efficiency of such algorithms applied to the estimation of the response of offshore wind turbines under combined wind and wave action still requires systematic analysis to select the optimal algorithm and apply it to the analysis of fatigue damage characteristics of offshore wind turbines.In this paper,the response characteristics of a monopile NREL-5MW wind turbine under combined wind and wave action are investigated and the accuracy and applicability of applying artificial intelligence algorithms to structural damage prediction is analyzed.The main research elements are as follows.(1)Based on the basic theory of wind and waves,harmonic superposition and fast Fourier inversion methods are used to generate environmental loads that can truly reflect the information of wind and wave fields.The joint wind and wave distribution functions are used to calculate the probability of occurrence of different operating conditions.Calculation of the tower base and blade load time courses and analysis of their time course and power spectrum characteristics to verify the correctness of the simulation results.Calculate the short-term fatigue damage at the base bottom as a training and validation database for algorithm prediction.(2)The mapping relationship between load input and fatigue output is established by three artificial intelligence algorithms,and the prediction effect of the models under different error judging indexes is considered,and the prediction capability of the three methods is analyzed from two perspectives of computational efficiency and computational accuracy.The influence of the number of training samples on the prediction results is explored to provide suggestions for the application scenarios of different algorithms.The results show that artificial intelligence algorithms can significantly improve the fatigue damage assessment efficiency of offshore wind turbines while ensuring a certain degree of accuracy,with the artificial neural network algorithm having the best results,requiring about 40 times higher computational efficiency and a maximum error of about 2.46%in total fatigue damage estimation compared to traditional time-domain calculations.The Gaussian process regression prediction results are less affected by the number of training samples,so when it is difficult to obtain more training data in engineering,Gaussian process regression is more suitable for calculating the fatigue damage of offshore wind turbines.(3)Based on the established damage prediction model,the fatigue characteristics of fixed offshore wind turbines under wind and wave loads are revealed.The fatigue damage characteristics of monopile offshore wind turbines are quickly analyzed by intelligent prediction algorithms,and it is found that for wind and wave effects alone,the structural damage caused by wind loads dominates:the direct superposition method and the sum-of-squares leveling method of wind and wave effects will underestimate the structural damage,while the combined method considering the slope of the SN curve will overestimate the structural damage.For combined wind and wave action,the smaller the wave period,the greater the damage,while for higher wave heights,the larger the wave period,the greater the damage;wind-induced damage plays a dominant role at low wave periods,while damage at high wave periods is mainly influenced by wave height.