具有优先级和机会约束的风力发电场实时优化

Abstract The article focuses on the efficient management of grid-connected wind farms, with the objective of optimizing production based on operational and economic criteria, while respecting short-term operational constraints and long-term energy production objectives. To achieve this, a deterministic Real Time Optimization (RTO) formulation based on stationary models is proposed and run hourly. This formulation aims to eliminate significant sources of uncertainty by using AutoRegressive Integrated Moving Average (ARIMA) models to predict the average wind speed for the short-term (one hour), while employing simple chance constraints to account for long-term variability in wind power production. These constraints are based on the inverse probability distribution of the producible power of the wind farms. The proposed problem presents multiple objectives with different levels of importance, leading to the comparison of two different methods for solving multi-objective problems: the weighted sum and the lexicographic method. This paper demonstrates that, although the weighted sum is a widely used method, it is unsuitable for the proposed formulation because it may not respect the relative importance of the objectives. In contrast, the lexicographic method ensures that the objectives are satisfied according to their importance. The proposed formulation has been applied to a case study involving 11 wind farms, evaluating different scenarios. It is shown that the proposed formulation outperforms other alternatives in meeting long-term energy production objectives. Additionally, it is demonstrated that the management solution is more efficient when long-term energy production objectives are applied to groups of wind farms rather than individual wind farms.