Machine Learning for Wind Energy Prediction

source: inhabitat.com article "Norway announces plans for Europe’s largest onshore wind farm"

Motivation

The produced power from a wind farm is varying with time. Reliable and accurate forecasting methods is required for a proper integration of this variable power in the energy system. This integration will be even more important in the future as the contribution from wind power in the total energy mix will continue to rise in many regions. Power forecasting methods is often a blend between deterministic models, like meso and micro scale flow models and statistical models, like Artificial Neural Networks ANN.

The objective is to validate power forecasting methods tailored towards the power forecasting of wind farms on a per turbine level. A blend of deterministic and statistical models will be used.

Goal

To develop and validate new forecasting methods for wind farms

What will I do?

* Survey of existing forecasting methods for wind power

* Develop machine learning-based new forecasting methods on a per turbine level

* Validate the developed algorithms and model in real dataset provided wind farm company

 

Supervisors

Yan Zhang, IFI, University of Oslo, Norway. Email: yanzhang[AT]ifi.uio.no

Arne Gravdahl, WindSim AS / Norwegian University of Life Sciences. Email: arne.gravdahl[AT]windsim.com

 

Publisert 22. nov. 2016 17:43 - Sist endret 22. nov. 2016 19:18

Veileder(e)

Omfang (studiepoeng)

60