Pierre-Julien Trombe

Probabilistic Forecasts of Wind Power Generation Accounting for Geographically Dispersed Information

Forecasts of wind power generation in their probabilistic form are a necessary input to decision-making problems for reliable and economic power systems operations in a smart grid context. Thanks to the wealth of spatially distributed data, also of high temporal resolution, such forecasts may be optimized by accounting for spatio-temporal effects that are so far merely considered. The way these effects may be included in relevant models is described for the case of both parametric and non-parametric approaches to generating probabilistic forecasts. The resulting predictions are evaluated on the real-world test case of a large offshore wind farm in Denmark (Nysted, 165 MW), where a portfolio of 19 other wind farms is seen as a set of geographically distributed sensors, for lead times between 15 minutes and 8 hours. Forecast improvements are shown to mainly come from the spatio-temporal correction of the first order moments of predictive densities. The best performing approach, based on adaptive quantile regression, using spatially corrected point forecasts as input, consistently outperforms the state-of-the-art benchmark based on local information only, by 1.5%-4.6%, depending upon the lead time.

Automatic Classification of Offshore Wind Regimes With Weather Radar Observations

Weather radar observations are called to play an important role in offshore wind energy. In particular, they can enable the monitoring of weather conditions in the vicinity of large-scale offshore wind farms and thereby notify the arrival of precipitation systems associated with severe wind fluctuations. The information they provide could then be integrated into an advanced prediction system for improving offshore wind power predictability and controllability. In this paper, we address the automatic classification of offshore wind regimes (i.e., wind fluctuations with specific frequency and amplitude) using reflectivity observations from a single weather radar system. A categorical sequence of most likely wind regimes is estimated from a wind speed time series by combining a Markov-Switching model and a global decoding technique, the Viterbi algorithm. In parallel, attributes of precipitation systems are extracted from weather radar images. These attributes describe the global intensity, spatial continuity and motion of precipitation echoes on the images. Finally, a CART classification tree is used to find the broad relationships between precipitation attributes and wind regimes.

Regime-based supervisory control to reduce power fluctuations from offshore wind power plants

Wind power fluctuations, especially offshore, can pose challenges in the secure and stable operation of the power system. In modern large offshore wind farms, there are supervisory controls designed to reduce the power fluctuations. Their operation is limited due to the fact that they imply loss of production, hence revenue for the wind farm operator. On the other hand, progresses in short term forecasting, together with the increasing use of probabilistic forecasting can help in achieving efficient power fluctuations reduction with minimum lost production. Here we present supervisory control concepts that consider different wind power regimes to derive control setpoints by using a Markov-Switching AutoRegressive model. We evaluate the performance versus measured data in terms of power ramp characteristics and energy efficiency.