宁夏本地化WRF辐射预报订正及光伏发电功率预测方法初探

Correction of Solar Radiation Forecast and Photovoltaic Power Prediction in Ningxia by Localization WRF Model

为了开展西北地区宁夏太阳能光伏发电气象预报服务,基于本地化WRF模式产品及当地光伏电站提供的发电功率资料,提出了EOF分析结合MOS预报的技术方法,进行模式辐射预报的统计订正研究,建立逐时光伏发电功率预测模型,得到较为理想的结果。通过EOF-MOS方法进行辐射预报订正,可使辐照度平均绝对百分比误差(MAPE)由原来24%左右降低到15%左右,模型预测发电功率MAPE稳定在22%左右。尤其是对于转折性天气趋势把握,具有较高的参考价值。

In order to carry out solar photovoltaic power generation forecast service in Ningxia of northwestern China,we used the combination of EOF analysis and MOS forecast method to correct the shortwave radiation WRF model based on the localized WRF model and power generation data from the local photovoltaic power station,and established an hourly photovoltaic power prediction model.Forecast revised by the EOF-MOS method could cause the mean absolute percentage error（MAPE） in solar irradiance forecast reduced from 24% to 15%,and the MAPE in photovoltaic power forecast be stable around 22%.The forecast method presents a good prediction result,especially for the forecast in frequently variable weather.