The actual load borne by a structure throughout its entire lifecycle, such as the wind load on fixed offshore wind turbines, is unknown. This leads to significant differences between the actual life of a structure and the fatigue analysis results during the design phase. Wind-load identification can effectively obtain the wind load borne by a fixed fan throughout its lifecycle. However, measurement noise and an ill-posed mathematical model affect the identification accuracy and stability. In this study, the optimal orientation of the sensor was determined to improve the signal-to-noise ratio according to the stress analysis of the tower structure. A successive addition method (SAM) was proposed to reduce the ill-posedness and dimensions of the mathematical model used for wind-load identification in the frequency domain by removing inefficient strain gauge locations and directions. A distributed successive addition method (DSAM) was developed to improve computational efficiency. The SAM and DSAM were described in detail using a specific case study of an offshore gravity wind turbine; a mathematical model was established through structural harmonic-response analysis. The spectrum superposition method and Dirlik distribution were used to analyse the fatigue damage of the wind-power tower. Several numerical examples were used to demonstrate the effectiveness of the SAM and DSAM. The simulation results demonstrated that the SAM and DSAM can completely eliminate the ill-posedness of mathematical models. In addition, the results of the structural fatigue analysis using the identified wind loads were extremely similar to the fatigue analysis results using the original wind loads, with an error of less than 3%.
Read full abstract