针对半片光伏组件电致发光（electroluminescence，EL）缺陷自动识别过程中训练用样本不足导致模型过拟合的问题,采用深度卷积生成对抗网络（deep convolutional generative adversarial networks，DCGANs）生成可控制属性的半片光伏组件EL图像，再采用多尺度结构相似性（multi-scale structural similarity，MS-SSIM）指标对生成的EL图像与拍摄的EL图像之间的相似程度进行了评估。评估结果得到，使用DCGANs生成的所有类型半片光伏组件的EL图像与拍摄的EL图像的MS-SSIM指标都大于0.55，大部分的MS-SSIM值在0.7附近。在分类模型的训练过程中，测试集准确率随着训练集中生成图像数量的增加而升高，当生成图像数量达到6000张时，测试集准确率达到97.92%。实验结果表明，采用DCGANs能够生成高质量且可控制属性的半片光伏组件EL图像，较好地解决因缺少训练样本而导致的模型过拟合问题。
Abstract
Aiming at the problem of model overfitting caused by insufficient training samples in the automatic identification process of electroluminescence (EL) defects for half-cut photovoltaic modules, the deep convolutional generative adversarial networks (DCGANs) were adopted to generate the half-cut photovoltaic module EL images with controllable attributes. The similarity between the generated EL images and the captured EL images was evaluated by using the multi-scale structural similarity (MS-SSIM) index. The evaluation results show that the MS-SSIM indexes of all types of EL images generated by DCGANs and captured EL images are greater than 0.55, and most of the MS-SSIM values are near 0.7. In the training process of the classification models, the accuracy of the test set increases with the increase of the number of images generated in the training set. When the number of generated images reaches 6 000, the accuracy of the test set reaches 97.92%. The experimental results show that the DCGANs can generate half-cut photovoltaic module EL images with high quality and controllable attributes, which can better solve the problem of model overfitting caused by the lack of training samples.
关键词
电致发光(EL) /
神经网络 /
多尺度结构相似性(MS-SSIM) /
深度卷积生成对抗网络(DCGANs)
{{custom_keyword}} /
Key words
deep convolutional generative adversarial networks (DCGANs) /
electroluminescence (EL) /
multi-scale structural similarity (MS-SSIM) /
neural network
{{custom_keyword}} /
基金
福建省市场监督管理局科技项目（FJMS2020024）；国家市场监督管理总局科技计划项目（2021MK050）
{{custom_fund}}
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
刘怀广, 刘安逸, 周诗洋, 等. 基于深度神经网络的太阳能电池组件缺陷检测算法研究[J]. 应用光学,2020,41(2):327-336.
周颖, 毛立, 张燕, 等. 改进CNN的太阳电池缺陷识别方法研究[J]. 太阳能学报, 2020, 41(12): 69-76.
刘怀广, 丁晚成, 黄千稳. 基于轻量化卷积神经网络的光伏电池片缺陷检测方法研究[J]. 应用光学,2022,43(1):87-94.
那峙雄, 樊涛, 孙涛, 等. 多损失融合的小样本光伏组件隐裂检测算法[J]. 计算机科学与探索,2022,16(2):458-467.
丛晓峰, 章军, 胡强. 基于多尺度空洞卷积的对抗去雾网络[J]. 应用光学,2020,41(6):1207-1213.
徐英, 谷雨, 彭冬亮, 等. 基于DRGAN和支持向量机的合成孔径雷达图像目标识别[J]. 光学精密工程,2020,28(3):727-735.
刘坤, 文熙, 黄闽茗, 等. 基于生成对抗网络的太阳能电池缺陷增强方法[J]. 浙江大学学报(工学版),2020,54(4):684-693.
周颖, 叶红, 王彤, 等. 基于多
LIU Huaiguang, LIU Anyi, ZHOU Shiyang, et al. Research on detection agorithm of solar cell componentdefects based on deep neural network[J]. Journal of Applied Optics,2020,41(2):327-336.
ZHOU Ying, MAO Li, ZHANG Yan, et al. Research on defect detection and classification for solar cells based on improved convolutional neural network[J]. Acta Energiae Solaris Sinica, 2020, 41(12): 69-76.
LIU Huaiguang, DING Wancheng, HUANG Qianwen. Defects detection method of photovoltaic cells based on lightweight convolutional neural network[J]. Journal of Applied Optics,2022,43(1):87-94.
NA Zhixiong, FAN Tao, SUN Tao, et al. Micro-cracks detection of solar cells based on few shot samples with multi-loss[J]. Journal of Frontiers of Computer Science & Technology,2022,16(2):458-467.
GOODFELLOW I J, POUGET-ABADIE J, MIRZA M, et al. Generative adversarial nets[C]//Proceedings of the 27th International Conference on Neural Information Processing Systems, Montreal, Dec 8-13, 2014. Cambridge: MIT Press, 2014: 26
{{custom_fnGroup.title_cn}}
脚注
{{custom_fn.content}}
