A mixed simulation methodology for long-term wind power with 4D fluctuation features clustering method and asymmetric fluctuations

Abstract

Long-term wind power time series (WPTS) simulation plays an important role in long-term grid planning. The study focuses on building a mixed simulation methodology for the long-term WPTS based on asymmetric fluctuations. First, we combine fixed threshold wavelet denoising with wavelet transform to get less noise and more detailed information. Second, a 4D fluctuation features clustering method (FFCM) is proposed to optimize the clustering results. In addition, based on multiple asymmetric function fitting models (AFFMs), an improved Markov Chain Monte Carlo (MCMC) method is proposed to prevent wind power ramp events. Numerical experimental results show that the average simulation error of simulated WPTS based on AFFMs is 3.02% lower than that of the symmetric function fitting model (SFFM). Therefore, the mixed simulation methodology can decrease the simulation error and improve the model accuracy in simulating long-term WPTS and provides a mathematical model for policy guidance on efficient use of wind resources.