风电场输出功率的多时段联合概率密度预测

Joint Probability Density Forecast for Wind Farm Output in Multi-time-interval

风电场输出功率波动性较强,难以精确预测,掌握其输出功率的分布规律对含有风电场的电力系统的运行决策具有重要意义。文中在分析风电场有功功率输出特性的基础上,提出了风电场输出功率多时段联合概率密度预测,利用风电场输出功率在时段间较强的相关性,估计其波动的幅度与速度特征,为系统运行提供更全面的决策信息。结合多元回归估计常条件相关—多元广义自回归条件异方差(CCC-MGARCH)模型与稀疏贝叶斯学习方法,给出了一种基于数值天气预报信息的风电场输出功率短期多时段联合概率密度预测方法。该方法依据CCC-MGARCH模型思想,将未来多个时段内风电场输出功率的联合概率密度预测问题分解为:风电场在各个时段内独立的输出功率概率密度预测子问题和时段间关联的输出功率预测误差相关系数矩阵估计子问题,利用稀疏贝叶斯学习方法在概率密度预测问题上的优势,形成预测效果好、计算效率高的风电场输出功率多时段联合概率密度预测方法。应用实例与分析说明了该方法的有效性。

Wind farm output power（WFOP） has a strong fluctuation and is difficult to precisely forecast.Knowing the distribution rules of WFOP is vital for the operation of power systems with wind farms.For this reason,by analyzing the characteristics of output power of wind farms,this paper proposes a joint probability density forecast method.According to close correlation of WFOP between time intervals,the range and rate characteristics of the fluctuation is estimated to provide more comprehensive information for power system operation decision-making.Based on constant conditional correlation-multivariate generalized auto regressive conditional heteroskedasticity（CCC-MGARCH） model and sparse Bayesian learning method,the paper proposes a short-term multi-time-interval joint probability density forecast approach for WFOP by using numerical weather prediction information.According to CCC-MGARCH model,the multi-time-interval joint probability density forecast for WFOP is divided into two parts： probability density forecast for WFOP at each single time interval and correlation coefficient matrix estimation for WFOP between the time intervals.Then,an effective and efficient multi-time-interval joint probability density forecast approach with the advantages of Sparse Bayesian Learning method at probabilistic density forecast is formed.Finally,an application example is given to prove the effectiveness of the proposed approach.