Wind:
Improving Statistical Wind Power Forecasting Methods

Windfarm

In the first phase of the research project, we surveyed existing wind power forecasting methodologies and models, and identified strengths and limitations of different approaches. Through a comprehensive state-of-the-art review, we developed a synthesis of aspects that are important from an end-user point of view for the selection and application of short-term wind power forecasting tools. We also identified areas of improvement in existing wind power forecasting methodologies. We are currently addressing critical issues from the state-of-the art review that may support the development of better statistical methods and tools for wind power forecasting, including

• Extracting more information from higher moments of the joint error probability distributions using concepts from Information Theoretic Learning
• Improving uncertainty estimation associated with forecasting errors
• Better representation of forecasting uncertainty for use in operational decision problems (e.g. unit commitment, scheduling, and wind power bidding)
• Development of an enhanced methodology for the prediction of ramping events

We are completing our wind power forecasting prototype and will thoroughly test and document the new algorithms developed as part of this. In particular, we are focusing on improvements in probabilistic forecasts and consistent scenario generation for use in operational decision problems for power system operators and wind farm owners. We will test our prototype using data from our industry partners.

We will also support the upcoming DOE/NOAA project on short-term wind power forecasting. Building on our ongoing research on wind power forecasting and its use in power system operations, we will assist in the evaluation of the impacts of the new meteorological measurements on forecasts of power production and benefits to electric power system operations.

In this research area, we partner with INESC Porto, a research institute in Portugal, Horizon Wind Energy, and the Midwest Independent System Operator.

Reports

Use of Wind Power Forecasting in Operational Decisions -- In this report, we analyze how wind power forecasting can serve as an efficient tool toward this end. We discuss the current status of wind power forecasting in U.S. electricity markets and develop several methodologies and modeling tools for the use of wind power forecasting in operational decisions, from the perspectives of the system operator as well as the wind power producer. In particular, we focus on the use of probabilistic forecasts in operational decisions. Read the report.

Development and Testing of Improved Statistical Wind Power Forecasting Methods -- In this report, we document improved statistical methods for wind power forecasting (WPF). First, we present the results of the application of information theoretic learning training criteria to wind power point forecasting. Second, we present novel time-adaptive kernel density forecast methods for characterizing WPF uncertainty, along with the corresponding case study results. Finally, a new method to predict and visualize ramp events is illustrated. Read the report.

More

• Read more about wind power forecasting
• Argonne press release
• Partner press release
• Science Daily article

December 2011