A Probabilistic-based Methodology for Wind Power Forecasting considering Generator Reliability

The massive increasing of wind generators in modern electrical grids is pushing power system operators to deal with the critical issues related to their intermittent and non-programmable power profiles, which sensible affect the effectiveness of critical power systems operation functions, such as security analysis and spinning reserve requirements. To deal with these challenging issues, a large number of forecasting models have been proposed in literature for predicting, on short and medium time horizons, the evolution of the main weather variables ruling the wind generators power production. Although these models allows system operators to mitigate the effects of wind power uncertainty, further, and more complex phenomena, should be analyzed in order to reliably predict the actual wind power profiles, such as the impacts of weather variables on generator operation state. To address this issue, in this paper a probabilistic model based on Markov chains is proposed to predict the generated power profiles considering the wind speed forecasting, and the expected generator operation state, i.e. alarm, fault and derated. Detailed experimental results obtained on a real case study are presented and discussed in order to demonstrate the effectiveness of the proposed methodology.