-
摘要:
传统的厂站一次接线图的绘制和管理主要依靠电网运行人员，费时费力且缺乏科学可校核的参考标准。提出了一种基于深度神经网络和数字图像处理相结合的厂站一次接线图的自动检测、识别和校核算法。首先，使用目标检测Faster R-CNN模型检测厂站接线图中的电器元件，并达到92%的检测准确率，同时使用端到端的文字检测识别模型识别厂站接线图中的文字信息，并达到94.2%的文字检测准确率和92%的文字识别准确率；然后，使用数字图像处理技术进行厂站接线图连接线、拓扑关系识别；最后，使用改进的VF2算法进行厂站一次接线图和人工维护的厂站一次接线图拓扑关系匹配校核，将拓扑数据抽象为无向图，通过轮廓序号得到元件的相对位置信息，根据改进的VF2算法得到2张图的匹配率，并通过匹配率与设定好的阈值来帮助核验，相比于节点遍历的匹配方法，核验准确率提高了37.5%。基于某供电公司提供的部分变电站的厂站一次接线图标注了接线图电器元件，贡献了一个小型接线图数据集。
Abstract:The drawing and management of the traditional primary wiring diagram of plant and station mainly depend on the operators of power grid, which wastes time and labor and lacks scientific and verifiable reference standards. Towards this problem, based on the deep neural networks and digital image processing, we propose an algorithm for automatic detection, recognition and verification. Specifically, Faster R-CNN is first adopted to detect the electrical components of wiring diagram with 92% detection accuracy. In the meantime, an end-to-end text detection and recognition model is used to recognize the text with 94.2% detection accuracy and 92% character recognition accuracy. Then we take advantage of digital image processing technique to identify the connection of wiring diagram and topological relation. Finally, improved graph matching algorithm VF2 is used to check the difference between the electronic and manually maintained diagrams. The topological data is abstracted into an undirected graph, and the relative position information of components is obtained through the outline number. Based on the improved VF2 algorithm, we can compute the matching rate of two graphs to help the verification. Compared with the matching method of node traversal, the verification accuracy can be improved by 37.5%. Based on the first wiring diagram of some substations provided by a power supply company, this paper marks the electrical components of wiring diagram and contributes a small wiring diagram dataset.
-
图 1 厂站一次接线图识别流程
Figure 1. Identification flowchart of primary wiring diagram of plant and station
图 2 厂站接线图多尺度融合检测算法
Figure 2. Multi-scale fusion detection algorithm for wiring diagram of plant and station
图 3 厂站接线图文字识别流程
Figure 3. Text recognition process of wiring diagram of plant and station
图 4 厂站接线图文字识别分割
Figure 4. Text recognition segmentation of wiring diagram of plant and station
表 1 电器元件识别实验结果对比
Table 1. Comparison of electrical component recognition experiment results
类别 检测准确率/% YOLOv3 Faster R-CNN Faster R-CNN融合算法 断路器 100 84 85.6 接地刀闸 92.7 90.5 96.6 接地 85.6 72.7 87.6 电抗 85.2 89.3 97.9 刀闸 95.3 86.7 94.8 变压器 96.9 100 100 避雷器 93.1 80 94.4 母线 87.5 85.9 97.2 手车开关 89.1 95.1 96 熔断器 93.9 92.7 93.3 电容 82.6 66.5 88.8 隔离手车1 87.2 0.72 76 隔离手车2 92.2 58.4 81.8 所变 94.6 86.1 88.9 电压互感器 96.4 88.6 85.7 电力电感器 75.4 83.3 94.4 忽略部件 80 66.7 80 平均 89.5 83.5 92 
下载: 导出CSV
表 2 文字识别实验结果对比
Table 2. Comparison of text recognition experiment results
检测算法 文本区域检测准确率/% 文本识别准确率/% Advanced EAST+Tesseract OCR 76.6 78 Attention-OCR 94.2 92
下载: 导出CSV
表 3 拓扑关系识别测试结果
Table 3. Test results of topological relation recognition
图片名称 正确的连接关系数 连接关系总数 准确率/% 10097 71 85 83.53 10099 115 119 96.64 10125 93 103 90.29 10129 67 80 83.75 20008 88 92 95.65 20009 68 104 65.38 20014 149 151 98.68
下载: 导出CSV
表 4 两种方法的匹配率结果
Table 4. Matching rate results of two methods
测试组别 匹配率/% 节点遍历方法 改进的VF2算法 第1组 27.97 55.95 第2组 37.50 51.61 第3组 0.74 54.74 第4组 62.07 63.49 第5组 24.42 76.67 第6组 9.33 92.00 第7组 93.24 92.86 第8组 0 71.69
下载: 导出CSV
-
[1] 刘建, 吴媛, 刘巩权. 配电馈线地理图到电气接线图的转换[J]. 电力系统自动化, 2005, 29(14): 73-77. https://www.cnki.com.cn/Article/CJFDTOTAL-DLXT200514016.htmLIU J, WU Y, LIU G Q. Conversion form GIS based feeder maps to electric diagrams[J]. Automation of Electrical Power Systems, 2005, 29(14): 73-77(in Chinese). https://www.cnki.com.cn/Article/CJFDTOTAL-DLXT200514016.htm [2] 沙树名, 林峰. 一种基于CIM的厂站接线图自动生成技术[J]. 电力系统自动化, 2008, 32(21): 68-71. doi: 10.3321/j.issn:1000-1026.2008.21.015SHA S M, LIN F. A technique of automatic generation of substation wiring diagram based on CIM[J]. Automation of Electric Power Systems, 2008, 32(21): 68-71(in Chinese). doi: 10.3321/j.issn:1000-1026.2008.21.015 [3] 苏运光, 罗俊, 熊浩, 等. 基于存量图形拓扑分析的厂站接线图自动生成[J]. 电力系统自动化, 2020, 44(3): 162-168. https://www.cnki.com.cn/Article/CJFDTOTAL-DLXT202003020.htmSU Y G, LUO J, XIONG H, et al. Automatic generation of plant and station wiring diagram based on stock graphic topology analysis[J]. Automation of Electric Power Systems, 2020, 44(3): 162-168(in Chinese). https://www.cnki.com.cn/Article/CJFDTOTAL-DLXT202003020.htm [4] KLUMP R, SCHOOLEY D, OVERBYE T. Advanced visualization platform for real-time power system operations[C]//Proceedings of the 14th Computation, 2002: 1-8. [5] 朱永利, 栗然, 刘艳. 电力系统厂站主接线图的自动生成[J]. 中国电力, 1999, 32(11): 56-58. https://www.cnki.com.cn/Article/CJFDTOTAL-ZGDL199911015.htmZHU Y L, LI R, LIU Y. Automatic generation of main wiring diagram of power system plant[J]. China Electric Power, 1999, 32(11): 56-58(in Chinese). https://www.cnki.com.cn/Article/CJFDTOTAL-ZGDL199911015.htm [6] 黄若航. 深度学习在热力管道工程设计上的研究和应用[D]. 成都: 电子科技大学, 2019.HUANG R H. Research and application of deep learning in thermal pipeline engineering design[D]. Chengdu: University of Electronic Science and Technology, 2019(in Chinese). [7] 尹思宇. 深度学习在变电站电力设备缺陷检测中的应用研究[D]. 成都: 电子科技大学, 2020.YIN S Y. Research on application of deep learning in defect detection of power equipment in substation[D]. Chengdu: University of Electronic Science and Technology, 2020(in Chinese). [8] REDMON J, FARHADI A. YOLOv3: An incremental improvement[EB/OL]. (2018-04-08)[2020-08-01]. https://arxiv.org/abs/1804.02767. [9] REDMON J, FARHADI A. YOLO9000: Better, faster, stronger[C]//IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway: IEEE Press, 2017: 17355115. [10] REN S, HE K, GIRSHICK R, et al. Faster R-CNN: Towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(6): 1137-1149. doi: 10.1109/TPAMI.2016.2577031 [11] ZHOU X, YAO C, WEN H, et al. EAST: An efficient and accurate scene text detector[C]//IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway: IEEE Press, 2017: 17355294. [12] SMITH R. An overview of the Tesseract OCR engine[C]//Ninth International Conference on Document Analysis and Recognition. Piscataway: IEEE Press, 2007: 9870037. [13] SIMONYAN K, ZISSERMAN A. Very deep convolutional networks for large-scale image recognition[EB/OL]. (2014-09-04)[2020-08-01]. https://arxiv.org/abs/1409.1556. [14] RONNEBERGER O, FISCHER P, BROX T. U-Net: Convolutional networks for biomedical image segmentation[EB/OL]. (2015-05-18)[2020-08-01]. https://arxiv.org/abs/1505.04597. [15] WOINA Z, GORBAN A N, LEE D S, et al. Attention-based extraction of structured information from street view imagery[C]//2017 14th IAPR International Conference on Document Analysis and Recognition (ICDAR). Piscataway: IEEE Press, 2017, 1: 844-850. [16] CAI Z, VASCONCELOS N. Cascade R-CNN: Delving into high quality object detection[EB/OL]. (2017-12-03)[2020-08-01]. https://arxiv.org/abs/1712.00726. [17] XU K, BA J, KIROS R, et al. Show, attend and tell: Neural image caption generation with visual attention[C]//Proceedings of the 32nd International Conference on Machine Learning, 2015, 37: 2048-2057. [18] SUZUKI S, BE K. Topological structural analysis of digitized binaryimages by border following[J]. Computer Vision Graphics and Image Processing, 1985, 30(1): 32-46. doi: 10.1016/0734-189X(85)90016-7 [19] CORDELLA L P, FOGGIA P, SANSONE C, et al. A (sub) graph is-omorphism algorithm for matching large graphs[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2004, 26(10): 1367-1372. doi: 10.1109/TPAMI.2004.75 [20] FOGGIA P, SANSONE C, VENTO M. A database of graphs for isomor-phism and sub-graph isomorphism benchmarking[C]//Proceedings of the 3rd IAPR-TC-15 International Workshop on Graph-based Representations, 2001: 176-188. [21] 李欣, 宋婷婷, 何险峰. 基于VF2算法的分子二维子结构检索[J]. 计算机与应用化学, 2007, 24(11): 1551-1554. doi: 10.3969/j.issn.1001-4160.2007.11.026LI X, SONG T T, HE X F. Molecular two-dimensional substructure retrieval based on VF2 algorithm[J]. Computer and Applied Chemistry, 2007, 24(11): 1551-1554(in Chinese). doi: 10.3969/j.issn.1001-4160.2007.11.026 [22] 楼昀恺, 王朝坤. 使用社区结构信息的子图匹配算法优化方法[J]. 计算机科学与探索, 2019, 13(1): 1-22. https://www.cnki.com.cn/Article/CJFDTOTAL-KXTS201901001.htmLOU Y K, WANG C K. Optimization method of subgraph matching algorithm using community structure information[J]. Computer Science and Exploration, 2019, 13(1): 1-22(in Chinese). https://www.cnki.com.cn/Article/CJFDTOTAL-KXTS201901001.htm [23] SHI B, YAO C, LIAO M, et al. ICDAR2017 competition on reading Chinese text in the wild (RCTW-17)[C]//2017 14th IAPR International Conference on Document Analysis and Recognition (ICDAR). Piscataway: IEEE Press, 2017, 1: 1429-1434. [24] CHRISTIAN S, SERHEY I, VINCENT V, et al. Inception-v4, inception-resnet and the impact of residual connections on learning[C]//Thirty-First AAAI Conference on Artifificial Intelligence, 2017: 1-12. -
下载:
点击查看大图
计量
- 文章访问数: 658
- HTML全文浏览量: 98
- PDF下载量: 152
- 被引次数: 0
