Abstract: To enhance the accuracy of wind speed simulations using S transform (ST), this paper proposed a high-precision data-driven approach based on multivariate empirical mode decomposition (MEMD) for nonstationary wind speed simulations in the field of structural wind engineering. A set of measured nonstationary wind speeds with multiple variables was selected. Intrinsic mode functions (IMFs) were obtained via MEMD and subjected to ST. The correlation between multiple variables was determined by employing a proper orthogonal decomposition (POD). Inverse ST (IST) was applied by incorporating random initial phases to generate simulated wind speeds. A time-frequency analysis was conducted to evaluate the proposed method. The results demonstrated that the proposed method effectively retained the energy characteristics of nonstationary wind speeds in the time domain. The amplitude distributions of the resulting ST coefficients closely resembled that of the measured wind speeds in the time-frequency domain. Quantitative comparisons of the average power spectrum confirmed the superior simulation accuracy of the proposed method compared with the ST simulation method based on the time-frequency power spectral density (TFPSD).

Key words: nonstationary wind speed, multivariate empirical mode decomposition (MEMD), S transform (ST), inverse S transform (IST), numerical simulation