基于最大类间方差法的天空图像云空识别模型

Cloud Identification Model for Sky Images Based on Otsu

准确的云空辨识对提高光伏发电功率超短期预测精度具有重要意义。首先针对由全天空成像仪采集的一系列天空图像进行预处理,然后提取样本图像的灰度矩阵,最后根据所提取的灰度矩阵建立基于最大类间方差法的天空图像云空识别模型。为验证本文提出的云空辨识模型的有效性,利用云南地区光伏电站中全天空成像仪(Total Sky Imager,TSI)采集的图像进行测试验证,与固定阈值法处理效果对比的结果表明,在复杂天气情况下面对不同分布特性的云团,本文提出的基于最大类间方差法的云空辨识模型更为准确、高效。

Accurate cloud identification is of great significance for increasing the accuracy of the ultra-short-term photovoltaic power forecast. Firstly,a series of sky images were pre-processed collected by Total Sky Imager（ TSI）. Secondly,the grayscale matrix of the sample image was extracted. Finally,based on the grayscale matrix,the sky image cloud identification model was established by using the Otsu method. In order to verify the effectiveness of the proposed method,we tested the image collected by TSI of PV power station in Yunnan Province and the results were compared with that of fixed threshold method. The results show that the cloud identification model based on Otsu in this paper can identify the sky and cloud more efficiently and accurately under complex weather conditions while facing different cloud clusters.