针对集中式光伏电站地形复杂、面积广、光伏组件故障识别较困难的情况,提出基于YOLOv8模型改进的光伏组件故障识别检测方法。基于Backbone结构采用渐进特征金字塔（AFPN）融合不同层级的图像提取多尺度信息,增强上下文信息的融合。通过在Neck结构添加无参数注意力机制（SimAM）,由能量函数推断出特征图中的三维注意力权重,轻量化地提高模型表征能力。取代每个池化层和每个跨步卷积层而建立SPD-Conv卷集神经网络,提高光伏组件图像中出现热斑、黑边和划痕等小目标特征的故障识别能力。实验结果表明,改进模型召回率和精确率分别达到78.7%和84.9%,平均精度mAP50和mAP50-95分别达到86%和57.9%,实现对光伏组件故障的识别与定位,验证该模型的正确性和有效性。
Abstract
Centralized photovoltaic power stations have complex terrain and wide areas, making it difficult to identify photovoltaic modules faults. An improved photovoltaic modules fault identification and detection method based on the YOLOv8 model is proposed. Based on the Backbone structure, the asymptotic feature pyramid network （AFPN） is used to fuse images at different levels to extract multi-scale information and enhance the fusion of contextual information. By adding a parameterless attention mechanism （SimAM） to the Neck structure, the three-dimensional attention weight in the feature map is inferred from the energy function, which lightweightly improves the model's representation ability. Instead of each pooling layer and each strided convolution layer, the SPD-Conv convolutional neural network is established to improve the fault identification ability of small target features such as hot spots, black edges and scratches in photovoltaic modules images. Experimental results show that the recall rate and accuracy rate of the improved model reach 78.7% and 84.9% respectively, and the average precision mAP50 and mAP50-95 reach 86% and 57.9% respectively. The identification and location of photovoltaic module faults are achieved and the correctness and effectiveness of the model is verified.
关键词
光伏组件 /
深度学习 /
目标检测 /
卷积神经网络 /
改进YOLOv8 /
注意力机制
Key words
PV modules /
deep learning /
object detection /
convolutional neural network /
improved YOLOv8 /
attention mechanism
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] 赵玲玲, 汪烨, 刘俊. 基于无人机与HSV空间的光伏电池板检测分析[J]. 红外技术, 2020, 42(10): 978-982.
ZHAO L L, WANG Y, LIU J.Detection and analysis of photovoltaic panels based on UAV and HSV space[J]. Infrared technology, 2020, 42(10): 978-982.
[2] 谢七月, 孙武彪, 申忠利, 等. 基于灰度聚类的光伏红外热斑图像分割方法[J]. 太阳能学报, 2023, 44(9): 117-124.
XIE Q Y, SUN W B, SHEN Z L, et al.Photovoltaic infrared hot spot image segmentation method based on gray clustering[J]. Acta energiae solaris sinica, 2023, 44(9): 117-124.
[3] 蒋兴群, 刘波, 宋力, 等. 基于改进YOLO-v3的风力机叶片表面损伤检测识别[J]. 太阳能学报, 2023, 44(3): 212-217.
JIANG X Q, LIU B, SONG L, et al.Surface damage detection and recognition of wind turbine blade based on improved YOLO-v3[J]. Acta energiae solaris sinica, 2023, 44(3): 212-217.
[4] ZHOU D Q, ZHAO W B, CHEN Y X, et al.Identification and localisation algorithm for sugarcane stem nodes by combining YOLOv3 and traditional methods of computer vision[J]. Sensors, 2022, 22(21): 8266.
[5] 赵伟达, 陈海文, 郭陆阳, 等. 基于YOLO-E与改进OCRNet图像分割的变电站仪表读数自适应识别方法[J]. 电力建设, 2023, 44(11): 75-85.
ZHAO W D, CHEN H W, GUO L Y, et al.Substation meter readings and dial information identification method based on YOLO-E and enhanced OCRNet image segmentation[J]. Electric power construction, 2023, 44(11): 75-85.
[6] 沈恒, 干宗良. 基于双阶段轻量YOLO的红外行人伪影检测算法[J]. 激光与红外, 2023, 53(9): 1426-1433.
SHEN H, GAN Z L.Forged shadow detection algorithm based on two-stage lightweight YOLO in infrared pedestrian[J]. Laser & infrared, 2023, 53(9): 1426-1433.
[7] 周颖, 王如意, 袁梓桐, 等. 一种高效双路径注意力太阳电池缺陷检测网络[J]. 太阳能学报, 2023, 44(4): 407-413.
ZHOU Y, WANG R Y, YUAN Z T, et al.An efficient dual-path attention solar cell defect detection network[J]. Acta energiae solaris sinica, 2023, 44(4): 407-413.
[8] QU S M, YANG X Y, ZHOU H F, et al.Improved YOLOv5-based for small traffic sign detection under complex weather[J]. Scientific reports, 2023, 13(1): 16219.
[9] 李婷, 孙渊. 基于改进轻量型YOLOv5的太阳能电池板缺陷检测[J]. 组合机床与自动化加工技术, 2023(11): 95-99, 106.
LI T, SUN Y.Defect detection of solar panels based on improved lightweight YOLOv5[J]. Modular machine tool & automatic manufacturing technique, 2023(11): 95-99, 106.
[10] HOSSEINY A, JAHANIRAD H.Hardware acceleration of YOLOv7-tiny using high-level synthesis tools[J]. Journal of real-time image processing, 2023, 20(4): 75.
[11] 苗荣慧, 李志伟, 武锦龙. 基于改进YOLO v7的轻量化樱桃番茄成熟度检测方法[J]. 农业机械学报, 2023, 54(10): 225-233.
MIAO R H, LI Z W, WU J L.Lightweight maturity detection of cherry tomato based on improved YOLO v7[J]. Transactions of the Chinese Society for Agricultural Machinery, 2023, 54(10): 225-233.
[12] 袁梦, 苏工兵, 王晶,等. 基于BCGS-YOLOv7 tiny的零件表面微小缺陷检测[J]. 制造技术与机床, 2023(10): 137-144.
YUAN M, SU G B, WANG J, et al.Part surface micro defect detection based on BCGS-YOLOv7 tiny[J]. Manufacturing technology & machine tool, 2023(10): 137-144.
[13] 周鑫, 李燕, 曾永辉, 等. 基于SARIMAX-SVR的光伏发电功率预测[J]. 电力系统及其自动化学报, 2024,36(5): 1-8.
ZHOU X, LI Y, ZENG Y H, et al.Forecasting of photovoltaic power generation based on SARIMAX-SVR[J]. Proceedings of the CSU-EPSA, 2024, 36(5): 1-8.
[14] 周颖, 颜毓泽, 陈海永, 等. 基于改进YOLOv8的光伏电池缺陷检测[J]. 激光与光电子学进展, 2024, 61(8): 245-255.
ZHOU Y, YAN Y Z, CHEN H Y, et al.Defect detection of photovoltaic cells based on improved YOLOv8[J]. Laser & optoelectronics progress, 2024, 61(8): 245-255.
基金
辽宁省应用基础研究计划（2022JH2/101300251）; 辽宁省教育厅科学研究项目（LJKZ1101）
PDF(1354 KB)
