Next generation forecasting tools for the optimal management of wind generation

Abstract

International audienceThis paper presents the objectives and an overview of the results obtained in the frame of the ANEMOS project on short-term wind power forecasting. The aim of the project is to develop accurate models that substantially outperform current state-of-the-art methods, for onshore and offshore wind power forecasting, exploiting both statistical and physical modeling approaches. The project focus on prediction horizons up to 48 hours ahead and investigates predictability of wind for higher horizons up to 7 days ahead useful i.e. for maintenance scheduling. Emphasis is given on the integration of high-resolution meteorological forecasts. Specific modules are also developed for on-line uncertainty and prediction risk estimation. An integrated software platform, 'ANEMOS', is developed to host the various models. This system is installed by several end-users for on-line operation at onshore and offshore wind farms for prediction at a local, regional and national scale. The applications include different terrain types and wind climates, on- and offshore cases, and interconnected or island grids

Improving the network infeed accuracy of non-dispatchable generators with energy storage devices

“…Forecasts can be calculated with various methods and approaches [13][14][15], whose forecasting technique is however not the subject of this investigation. The question addressed here is merely how a forecast error influences the ability of the system to meet the planned network infeed and to what extent a storage device can compensate forecast errors.…”

“…2 and repeatedly calculates the resulting amounts of surplus and insufficient energy for different values of x 1 to x n until the set is found minimising the target function in (15). The only required constraint concerns the relative values of the four variables to satisfy x 1 ≤ x 2 ≤ · · · ≤ x n , thus guaranteeing a continuous parametrisation.…”

A hybrid approach to balance the variability and intermittency of renewable generation

“…Predictions of available wind generation from methods as described in [13], [14] allow for preventive actions such as fully charging storage devices beforehand and/or slowly ramping up additional backup generation when a drop in wind generation is predicted. Hence, in this paper it is assumed that wind predictions with a resolution of 5 minutes are available.…”

Short-term wind power prediction based on combined grey-Markov model

“…Unexpected variations of a wind farm output may increase operating costs for the electricity system by increasing requirements of primary reserves, as well as increase potential risks to the reliability of electricity supply. In order to schedule the spinning reserve capacity and manage the grid operation, grid operators mainly using persistence-type methods to predict changes of the wind power production [3]- [4]. However, the wind production variations cannot be predicted accurately by using persistence-type methods, it is necessary to search better forecasting methods.…”