Wind Speed Prediction for Wind Farm Based on Clayton Copula Function and Deep Learning Fusion

摘要

Complex spatio-temporal dependencies exist between wind speed (WS) of various wind turbines (WT) in wind farm (WF), which brings great difficulties to obtain accurate WS-prediction (WSP) results. Analyzing the spatio-temporal dependency of WS, clarifying the mutual influence between WT, and constructing high-precision wind- speed prediction models are still problems to be solved. For this reason, a WSP model combining long and short-term memory (LSTM) network and Copula function is proposed to realize WSP based on the analysis of the wind-speed spatio-temporal dependency. The effectiveness of the method is verified by using the measured wind-speed data of the WF. The method effectively solves the problem of windspeed spatio-temporal dependency analysis and significantly improves the WSP accuracy.