Spatio-Temporal Modelling of Wind Power Ramps in Alberta
Date
2024-09-27
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The goal of this thesis is to model wind power ramps using a three-state Markov chain. The ramp detection technique employed is known as L1-SW in the literature. Within the Markov chain, the states’ transition probabilities are governed by a Gaussian process with a separable spatiotemporal covariance function, designed to capture the space-time dependencies across wind farms. The three states of the Markov chain are ramp up (+1), ramp down (-1), and non-ramp interval (0). The parameters of this model are estimated using a Bayesian inference framework, specifically employing no U-Turn Sampler (NUTS), which is a Hamiltonian Monte Carlo (HMC) Method. The inference procedure is implemented using RStan, an interface for working with Stan in R. The results demonstrate that our model effectively captures the properties of wind power ramps. This model is then extended to predict the ramping behavior of future, not-yet-established wind farms.
Description
Keywords
Wind power, Wind power ramps, Markov chain, Gaussian process, Spatio-temporal Process, Covariance function, Hamiltonian Monte Carlo, No U-Turn Sampler, RStan, Bayesian inference, Wind farm, HMC
Citation
Mahmoudi Gharaie, M. (2024). Spatio-temporal modelling of wind power ramps in Alberta (Master's thesis, University of Calgary, Calgary, Canada). Retrieved from https://prism.ucalgary.ca.