Wind Power Forecasts in Power System Operation and Electricity Markets
The integration of electrical energy from renewable sources, in particular wind power, into the power system and electricity market operation has motivated research of new decision-making methods.
Furthermore, in the last twenty years, research has been conducted for developing wind power forecasting (WPF) algorithms. Several authors explored the idea of including information from Numerical Weather Prediction (NWP) systems in WPF algorithms, aiming to produce forecasts with acceptable accuracy several days-ahead. The subsequent step was to develop algorithms for wind power uncertainty forecast. Two examples of statistical algorithms are quantile regression and conditional kernel density forecast. In parallel, NWP systems were also explored to produce uncertainty forecasts represented by ensemble of meteorological predictions.
Presently, one of the new research priorities is a fully integration of wind power forecasts and information on their uncertainties in the algorithms of the power system operational management tools, as well as in bidding algorithms for the electricity market. Examples of problems that require WPF are: setting the operating reserves requirements, power flow calculations, unit commitment and deriving the optimal selling bids for the electricity market.
My research in this topic comprises the following aspects:
Furthermore, in the last twenty years, research has been conducted for developing wind power forecasting (WPF) algorithms. Several authors explored the idea of including information from Numerical Weather Prediction (NWP) systems in WPF algorithms, aiming to produce forecasts with acceptable accuracy several days-ahead. The subsequent step was to develop algorithms for wind power uncertainty forecast. Two examples of statistical algorithms are quantile regression and conditional kernel density forecast. In parallel, NWP systems were also explored to produce uncertainty forecasts represented by ensemble of meteorological predictions.
Presently, one of the new research priorities is a fully integration of wind power forecasts and information on their uncertainties in the algorithms of the power system operational management tools, as well as in bidding algorithms for the electricity market. Examples of problems that require WPF are: setting the operating reserves requirements, power flow calculations, unit commitment and deriving the optimal selling bids for the electricity market.
My research in this topic comprises the following aspects:
- Use of concepts from Information Theoretic Learning for training neural networks. This avoids the assumption of Gaussian forecast errors
- Time-adaptive conditional kernel density forecast models for probabilistic WPF
- Decision-aid tool for setting the operating reserve requirements, using probabilistic WPF as input
- Decision-aid tool for performing power flow calculations, modelling uncertainties with fuzzy sets
- Bidding algorithms based on utility theory for deriving selling bids for wind farms owners
- Stochastic unit commitment with wind power generation