Wind power scenario generation for multiple wind farms based on atmospheric dynamic model

The wind power scenarios of multiple wind farms with temporal and spatial correlations play an important role for day-ahead or hours-ahead stochastic generation scheduling problems with significant wind power penetration. In this paper, we proposed a physical mechanism based analytical model to generate temporal and spatial correlated scenarios of wind power generation for multiple wind farms. A stochastic dynamic system was established on top of the atmospheric dynamic equations and wind speed downscaling equations to describe the relationship between the atmospheric and near-surface wind field for all wind farms. Combing with wind speed measurements, the extended Kalman filter was applied to estimate the system state and to forecast the joint probability density function (PDF) of wind speeds of multiple wind farms. Based on this joint PDF, the spatial and temporal correlated wind power scenarios of the wind farms were then generated through the procedures of Monte Carlo simulation, wind power conversion and scenario reduction. In case studies, the proposed method were tested and compared with Gaussian Copula based methods based on data of 4 wind farms in the State of Missouri in USA. The evaluation results verify the effectiveness of the proposed method.