Spatial-Temporal Gaussian Processes for Wind Power Generation

Yilan Luo (Mathematics and Statistics, University of Calgary) is sharing her Master's research work with us.

Abstract:

Wind is one of the renewable and green energy sources contributing to Alberta's energy program and economics. We are interested in the spatio-temporal features of wind power generation. A Gaussian process is utilized to model this feature.

We studied various covariance function models (separable, non-separable symmetric, and non-symmetric) by following Gneiting et al. (2006) and measured how good they are by comparison. The covariance function is an important characteristic of a spatio-temporal process, as it describes the association of random variables in space and time.

A simple Kriging technique is used to demonstrate the performance of models when forecasting the future wind generation for both an existing wind farm and a new farm. RMSE and MAE are used to measure the accuracy of prediction. Finally, we are sharing our experiences on the choice of the best practical scenario for new wind farms based on the aggregate wind power generation.

Hope to see you there. The event starts from 6 pm with pizza and presentation starts at 6:15 sharp.

Our sli.do channel is #V613 - go to https://sli.do/ and enter V613 for the channel. You will then be able to post silent questions during the presentations.