Simplified Method to Derive the Kalman Filter Covariance Matrices to Predict Wind Speeds from a NWP Model

Abstract

This paper espouses a simplified approach to predict wind speed 1 hour ahead for a wind turbine located on the Cork Institute of Technology (CIT) college campus by utilising a Kalman Filter to predict the bias between a campus based turbine and the output from a Numerical Weather Prediction (NWP) model for Cork Airport. Furthermore, this paper investigates the optimum number of samples required ( n ) in a fixed sampling interval process to derive the covariance matrix of the system equation Qt and the covariance matrix of the observation equation Rt . The main contents of this paper include wind speed analysis, state space analysis and Kalman Filtering application to Numerical Weather Prediction (NWP) data for wind speed prediction.