Unsupervised statistical estimation of offshore wind turbine vibration for structural damage detection under varying environmental conditions. (1st December 2022)

• Record Type: Journal Article
• Authors: Guo, Jianxun; Ji, Xiang; Song, Hong; Chang, Shuang; Liu, Fushun
• Is Part Of: Engineering structures. Volume 272(2022)
• Publication Date: 2022-12-01
• Subjects: Unsupervised statistical analysis -- Information entropy -- Analytic network process -- Structural damage detection -- Offshore wind turbine; Structural engineering -- Periodicals; Structural analysis (Engineering) -- Periodicals; Construction, Technique de la -- Périodiques; Génie parasismique -- Périodiques; Pression du vent -- Périodiques; Earthquake engineering; Structural engineering; Wind-pressure; Periodicals; 624.105
• DOI: 10.1016/j.engstruct.2022.115005
• Languages: English
• ISSNs: 0141-0296

Abstract

Abstract: Assessing the structural health condition of offshore wind turbines (OWTs) is necessary to ensure their later stages of service. A widely adopted strategy is to identify the damage by a significant deviation between the serviced structural responses and data obtained from undamaged structures. However, dynamic responses of OWTs show high dependency on ambient environments. To evaluate the variabilities of structural responses, in this paper an unsupervised statistical estimation method is proposed based on the integration of data bin processing and interval correlation techniques to eliminate the effects of environmental and operational conditions on evaluation results of a sensor, called a single indicator. Moreover, the improved network analysis model using information entropy has been established to consider the interactions of the multiple indicators for the valuation of the predicted results. As compared with the traditional methods, the proposed method combination between environmental variables clustering analysis and interval correlation could has demonstrated better robustness and higher computational efficiency using the developed probabilistic model. To validate the correctness and robustness of the proposed method, numerical simulations of a 5 MW monopile OWT by OpenFAST have been carried out. Based on the stiffness loss between 4% and 16% in natural frequencies and ten different wind speeds in the range of 5 to 23 m/s, four scenarios have been investigated to evaluate the sensitivity of damage detection. Results have shown that the proposed method has the ability to effectively and robustly assess the structural damage condition under different wind speed conditions. Furthermore, field test data from a 4 MW monopile OWT has been utilized to demonstrate the feasibility of the established statistical estimation method. It has been noted that the false alarm number of multiple indicators has been reduced to 0 from 3 in the case where there is no consideration of environmental condition classification and this evaluation result has well agreed with the data obtained under the actual working state of the OWT. Highlights: A statistical analysis method has been proposed to assess the performance of offshore wind turbine under varying environmental conditions. The analytic network algorithm based on information entropy has been implemented in the probabilistic model for the improved robustness. Numerical examples and the field data have been used to verify the proposed method.