A comprehensive error evaluation method for wind power forecasting and its application

Wind power forecasting evaluation indexes reflect the accuracy and reliability of prediction models, being an important basis for positioning and improving errors. The current evaluation indexes are numerous and mixed with each other, lacking a clear layer-by-layer classification, and their physical meanings are mostly on the statistical level and cannot reflect the interaction with the power grid. A comprehensive error evaluation method for wind power forecasting is established. Taking time scale as the main criterion, the evaluation method covers the mathematical and applied characteristics of errors. The former is organized from multiple dimensions such as expectation, extreme value and distribution, and diversified types such as point and probability prediction separately. The latter considers the impact of errors on power quality, day-to-day scheduling, and system planning in various stages. Finally, the data of an actual operating wind farm are taken as an example to illustrate the application and effectiveness of the evaluation method.