在当前地基云图分类任务中,存在识别准确率低等问题。为了提高云分类的精度,有效融合深度可分离卷积、注意力机制和残差结构的特点,构建DAR-CapsNet地基云图分类模型。首先,收集整理美国国家新能源实验室公开数据库中的地基云图,构建云分类数据库;然后,对所提出的DAR-CapsNet分类模型进行训练优化;最后,在不同数据集上验证所提出的分类模型性能。实验结果表明所提出的DAR-CapsNet分类模型,分类准确率高达95.50%,优于现有公开分类方法,且在不同数据集上具有较好的泛化性能。
Abstract
In the current ground-based cloud image classification task, there are problems such as low recognition accuracy. In order to improve the accuracy of cloud classification, the DAR-CapsNet classification model for ground-based cloud images has been constructed by effectively integrating the features of depthwise separable convolution, attention mechanism and residual structure. Firstly, the ground-based cloud images were collected from the public database of the National New Energy Laboratory of the United States to build a cloud classification database; then, the proposed DAR-CapsNet classification model was trained and optimized; finally, experiments were conducted on different datasets to verify the performance of the proposed classification model. The experimental results show that the classification accuracy of the DAR-CapsNet model is as high as 95.50%, which is better than some published classification models, and the DAR-CapsNet model has better generalization performance on different datasets.
关键词
光伏发电 /
气象云 /
图像分类 /
卷积神经网络 /
机器学习
Key words
photovoltaic power generation /
clouds /
image classification /
convolutional neural networks /
machine learning
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] 过奕任, 张梦叶, 朱婷婷, 等. 基于网络集成的太阳全辐射超短期混合预测方法[J]. 工业控制计算机, 2020, 33(9): 35-36, 39.
GUO Y R, ZHANG M Y, ZHU T T, et al.A hybrid model for very short-term forecast of global horizontal irradiance based on network ensemble[J]. Industrial control computer, 2020, 33(9): 35-36, 39.
[2] 倪超, 王聪, 朱婷婷, 等. 基于CNN-Bi-LSTM的太阳辐照度超短期预测[J]. 太阳能学报, 2022, 43(3): 197-202.
NI C, WANG C, ZHU T T, et al.Super-short-term forecast of solar irradiance based on CNN-Bi-LSTM[J]. Acta energiae solaris sinica, 2022, 43(3): 197-202.
[3] 张之正, 冯径, 徐志康, 等. 基于Android平台的地基云状识别系统[J]. 自动化技术与应用, 2019, 38(11): 127-131, 146.
ZHANG Z Z, FENG J, XU Z K, et al.Ground nephogram recognition system based on android platform[J]. Techniques of automation and applications, 2019, 38(11): 127-131, 146.
[4] 杨秋良, 王钰, 杨杏丽, 等. 基于互信息F统计量特征选择技术的地基气象云图分类[J]. 计算机与现代化, 2021(2): 18-23,29.
YANG Q L, WANG Y, YANG X L, et al.Classification of ground-based meteorological cloud images based on feature selection technique of mutual information F-statistics[J]. Computer and modernization, 2021(2): 18-23, 29.
[5] 李林, 刘青山, 夏旻. 基于极限学习机的地基可见光云图的云分类[J]. 大气科学学报, 2015, 38(5): 678-685.
LI L, LIU Q S, XIA M.Cloud classification based on ground-based vision cloud image using extreme learning machine[J]. Transactions of atmospheric sciences, 2015, 38(5): 678-685.
[6] 朱彪, 杨俊, 吕伟涛, 等. 基于KNN的地基可见光云图分类方法[J]. 应用气象学报, 2012, 23(6): 721-728.
ZHU B, YANG J, LYU W T, et al.Ground-based visible cloud image classification method based on KNN algorithm[J]. Journal of applied meteorological science, 2012, 23(6): 721-728.
[7] HEINLE A, MACKE A, SRIVASTAV A.Automatic cloud classification of whole sky images[J]. Atmospheric measurement techniques, 2010, 3(3): 557-567.
[8] KLIANGSUWAN T, HEEDNACRAM A.Feature extraction techniques for ground-based cloud type classification[J]. Expert systems with applications, 2015, 42(21): 8294-8303.
[9] KAZANTZIDIS A, TZOUMANIKAS P, BAIS A F, et al.Cloud detection and classification with the use of whole-sky ground-based images[J]. Atmospheric research, 2012, 113: 80-88.
[10] 王敏, 庄志豪, 王康, 等. 地基云图卷积神经网络自动分类的训练过程[J]. 信息技术, 2021, 45(10): 26-30.
WANG M, ZHUANG Z H, WANG K, et al.Training process of convolution neural network in automatic classification of ground-based cloud image[J]. Information technology, 2021,45(10): 26-30.
[11] ZHANG J L, LIU P, ZHANG F, et al.CloudNet: ground-based cloud classification with deep convolutional neural network[J]. Geophysical research letters, 2018, 45(16): 8665-8672.
[12] 贾克斌, 张亮, 刘鹏宇, 等. 基于轻量级神经网络的地基云图识别[J]. 北京工业大学学报, 2021, 47(5): 489-499.
JIA K B, ZHANG L, LIU P Y, et al.Ground-based cloud recognition based on lightweight neural network[J]. Journal of Beijing University of Technology, 2021, 47(5): 489-499.
[13] 崔顺, 许允飞, 苏丽颖, 等. 基于卷积神经网络的全天空地基云图分类研究[J]. 天文研究与技术, 2019, 16(2): 225-235.
CUI S, XU Y F, SU L Y, et al.Classification of all-sky camera data based on convolutional neural network[J]. Astronomical research & technology, 2019, 16(2): 225-235.
[14] 段向军, 王敏. 迁移深度学习地基云图自动识别的网络微调学习过程[J]. 电子器件, 2020, 43(6): 1257-1261.
DUAN X J, WANG M.An ground cloud image recognition method using AlexNet convolution neural network[J]. Chinese journal of electron devices, 2020, 43(6): 1257-1261.
[15] SABOUR S, FROSST N, HINTON G E.Dynamic routing between Capsules[J]. Advances in neural information processing systems, 30: 3859-3869, 2017.
[16] MAZZIA V, SALVETTI F, CHIABERGE M.Efficient-CapsNet: capsule network with self-attention routing[J]. Scientific reports, 2021, 11: 14634.
[17] STOFFEL T, ANDREAS A.NREL solar radiation research laboratory (SRRL): baseline measurement system (BMS); golden, Colorado (data)[R]. No.NREL/DA-5500-56488, 1981.
[18] DEV S, LEE Y H, WINKLER S.Categorization of cloud image patches using an improved texton-based approach[C]//2015 IEEE International Conference on Image Processing (ICIP). Quebec City, Canada: IEEE, 2015: 422-426.
基金
国家自然科学基金青年项目（62006120）
PDF(2099 KB)
