A Wind Power Forecasting Method Using Causal Inference

Most wind power forecasting is based on neural networks (NN) techniques. In this work, we use a causal inference approach in wind power forecasting techniques. First, the results from numerical weather prediction (NWP) are analyzed using potential-outcomes models from causal models. Then, based on the results of the causal discovery, the meteorological data obtained from NWP are judged and filtered by the variables that have a causal effect on the wind power output. Finally, the strength of the causal effect of each variable will be quantified and used as the basis for a new algorithm to optimize the wind power output forecasting method. Case study is a forecast analysis based on one year's NWP data from a single wind power plant.