中长期风电功率的多气象变量模型组合预测方法

Medium- or Long-Term Wind Power Prediction With Combined Models of Meteorological Multi-Variables

为了预防和控制危害性风电功率事件,提出多气象变量模型的组合预测方法,以实现中长期高精度风电功率预测。该方法利用数值天气预报提供的气象数据预测长期风电趋势,同时局部采用多变量模型改善预测精度。为了保证多变量模型的有效性,首先采用Granger因果检测法筛选出对风电功率预测有效的气象变量。其次,针对不同气象变量进行数据结构分析,并根据其动力学特性单独建立合适的预测模型,然后采用线性或非线性机制对不同气象变量预测结果进行组合,完成组合预测模型的建立。最后,通过对实例数据仿真,实现了中长期风电功率预测,并结合误差分析验证了组合预测模型的有效性,且预测结果为后续中长期风电功率事件分析提供了基础。

To control harmful wind power events, this paper proposed a combined model of different meteorological variables to realize wind power prediction with high accuracy. This method applies meteorological data from NWP to predict long-term trend of wind power, then utilizes multi-variable model locally to improve its performance. In order to guarantee validity of multi-variable model, Granger test method is used to screen valid meteorological variables for wind power prediction. Then, analyzing data structure of different variables, each variable is used to establish a prediction model respectively based on dynamic feature. To complete final model, linear and nonlinear regime are utilized to combine those prediction results of meteorological variables. Finally, long-term wind power prediction is completed based on test data, and validity of combined model is also verified with error indicator analysis.