Multi-information fusion welding defect identification combining neighborhood rough set and optimized SVM
-
摘要:
针对多传感器信息融合焊接过程产生的“大数据”，将邻域粗糙集 (neighborhood rough set, NRS)与优化支持向量机 (support vector machine, SVM)相结合，提出一种多信息融合焊接缺陷识别方法，以特征重要性为启发信息构造基于NRS的快速约简算法，利用野狗优化算法 (dingo optimization algorithm, DOA)选取SVM的关键参数，通过试验获取熔池图像、焊接电流和振动信号等焊接信息，采用特征融合与NRS约简生成精简的数据集，载入 DOA-SVM进行优化训练后建立焊接缺陷识别模型，设计多组试验对该方法进行对比验证. 结果表明，模型对 6种焊缝质量类别识别准确率为98.03%，且训练和预测时间短、泛化能力强，能满足焊接质量在线检测的要求.
Abstract:A multi-information fusion welding defect recognition method is proposed by combining NRS with optimized SVM to address the “big data” generated during the multi-sensor information fusion welding process. A fast reduction algorithm based on NRS was constructed using feature importance as inspiration information, and the DOA was used to select the key parameters of SVM. Through welding experiments, welding quality information such as melt pool images, welding currents, and vibration signals was obtained. A simplified dataset was generated using feature fusion and NRS reduction, and then loaded into DOA-SVM for optimization training to establish a welding defect recognition model. Multiple sets of experiments were designed to compare and verify the method. The results showed that the model achieved an accuracy of 98.03% in identifying six types of weld quality categories, with short training and prediction time and strong generalization ability, which could meet the requirements of online welding quality detection.
-
-
![]()
图 3 多信息融合焊接过程知识建模的基本流程
Figure 3. Basic flow of welding process knowledge modeling based on multi-information fusion
![]()
图 5 各质量类型焊缝外观与熔池图像轮廓
Figure 5. Appearance and pool contour images of welds of different quality types. (a) weld appearance; (b) geometric shape of molten pool
![]()
图 6 各焊接质量类型的灰度等级
Figure 6. Gray scale of each welding quality type. (a) 01; (b) 02; (c) 03; (d) 04; (e) 05; (f) 06
![]()
图 8 各质量类型电流信号时域特征
Figure 8. Time domain characteristics of current signals for various quality types. (a) standard deviation distribution; (b) mean distribution; (c) root mean square distribution; (d) peak to peak distribution; (e) peak factordistribution; (f) shape factordistribution; (g) skewness distribution; (h) kurtosis distribution
![]()
图 9 3种质量状态下焊接振动信号时域特征
Figure 9. Time domain characteristics of welding vibration signals under three different quality states. (a) mean distribution; (b) root mean square distribution; (c) root mean square amplitude distribution
![]()
图 10 各质量状态下焊接振动信号的3个频域特征
Figure 10. Three frequency domain characteristics of welding vibration signals under different quality states. (a) mean frequency; (b) standard deviation; (c) skewness
表 1 焊接多源信息的数据集
Table 1 Data set of welding multi-source information
编号 类型 数据组G/个 01 气孔 335 02 未焊透 340 03 烧穿 366 04 未焊满 359 05 良好 358 06 焊偏 162 
下载: 导出CSV
表 2 各信息源未约简与约简模型的分类质量
Table 2 Classification quality of unreduced and reduced models for each information source
信息源 未约简特征数 准确率
Acc(%)约简后特征数 准确率
Acc(%)电流信号 8 80.33 7 79.33 振动信号 87 78.85 40 79.02 熔池图像 139 94.22 13 94.30 多源信息 234 96.80 17 96.88
下载: 导出CSV
表 3 不同模型分类准确率对比(%)
Table 3 Comparisons of classification accuracy of different models
CART SVM DOA-SVM 未约简 95.81 96.80 97.79 约简后 94.01 96.88 98.03
下载: 导出CSV
表 4 约简的DOA-SVM模型试验结果
Table 4 Experimental results of reduced DOA-SVM model
编号 类型 精确度P 召回率R F1分数F 01 气孔 1.000 0.969 0.984 02 未焊透 0.937 1.000 0.967 03 烧穿 1.000 1.000 1.000 04 未焊满 0.989 1.000 0.994 05 良好 1.000 0.957 0.978 06 焊偏 1.000 1.000 1.000
下载: 导出CSV
-
[1] 李汝跃. 大型机械焊接工艺优化与参数控制研究[J]. 中国机械, 2023(20): 51 − 54. LI Ruyue. Research on optimization and parameter control of welding process for large machinery[J]. Machine China, 2023(20): 51 − 54.
[2] 刘鑫, 赵丽丽. 智能化机器人焊接技术研究进展[J]. 科学家, 2017, 5(8): 19 − 20. LIU Xin, ZHAO Lili. Research progress in intelligent robot welding technology[J]. Scientist, 2017, 5(8): 19 − 20.
[3] 张琨, 邹宗轩, 刘烨, 等. 铝合金TIG焊接熔池状态多传感器数据协同感知算法[J]. 焊接学报, 2022, 43(3): 50 − 55. ZHANG Kun, ZOU Zongxuan, LIU Ye, et al. Multi sensor data collaborative perception algorithm for aluminum alloy TIG welding melt pool status[J]. Transactions of the China Welding Institution, 2022, 43(3): 50 − 55.
[4] 于鹏. 铝合金脉冲多重控制GMAW焊接质量在线监测研究 [D]. 长春: 吉林大学, 2020. YU Peng. Research on online monitoring of aluminum alloy pulse multiple control GMAW welding quality [D]. Changchun: Jilin University, 2020.
[5] 张志芬. 基于多信息融合的铝合金脉冲GTAW过程焊接缺陷特征提取研究[D]. 上海: 上海交通大学, 2015. ZHANG Zhifen. Research on feature extraction of welding defects in aluminum alloy pulse GTAW process based on multiple information fusion [D]. Shanghai: Shanghai Jiao Tong University, 2015.
[6] 朱焘. K-TIG 焊接焊缝偏差与熔透状态识别的研究[D]. 广州:华南理工大学, 2019. ZHU Tao. Research on deviation and penetration status identification of K-TIG welding seams [D]. Guangzhou: South China University of Technology, 2019.
[7] HU Q H, YU D R, XIE Z X. Neighborhood classifiers[J]. Expert systems with applications, 2008, 34(2): 866 − 876. doi: 10.1016/j.eswa.2006.10.043
[8] 张迪. 邻域粗糙集模型的拓展研究 [D]. 北京: 北京邮电大学, 2023. ZHANG Di. Research on the extension of neighborhood rough set models [D]. Beijing: Beijing University of Posts and Telecommunications, 2023.
[9] PAWLAK Z. Rough sets: Theoretical aspects of reasoning about data [M]. Dordrecht: Kluwer Academic Publishing, 1991.
[10] ZHANG X, LIU C, XUE L, et al. Simultaneous feature selection and SVM parameter by using artificial bee colony algorithm [C]//Proceedings of the 2022 6th International Conference On Electronic Information Technology and Computer Engineering, 2022: 1737 − 1745.
[11] PERAZA-VÁZQUEZ H, PEÑA-DELGADO A F, ECHAVARRÍA-CASTILLO G, et al. A bio-inspired method for engineering design optimization inspired by dingoes hunting strategies[J]. Mathematical Problems in Engineering, 2021, 2021: 1 − 19.
[12] 冯林. 基于粗糙集理论的不确定信息处理与知识获取方法研究 [D]. 成都: 西南交通大学, 2008. FENG Lin. Research on uncertain information processing and knowledge acquisition methods based on rough set theory [D]. Chengdu: Southwest Jiaotong University, 2008
[13] 胡清华, 于达仁, 谢宗霞. 基于邻域粒化和粗糙逼近的数值属性约简[J]. 软件学报, 2008, 19(3): 640 − 649. doi: 10.3724/SP.J.1001.2008.00640 HU Qinghua, YU Daren, XIE Zongxia. Numerical attribute approximation based on neighborhood granularity and rough approximation[J]. Journal of Software, 2008, 19(3): 640 − 649. doi: 10.3724/SP.J.1001.2008.00640
[14] ALMAZÁN Covarrubias J H, PERAZA Vázquez H, PEÑA Delgado A F, et al. An improved dingo optimization algorithm applied to SHE-PWM modulation strategy[J]. Applied Sciences, 2022, 12(3): 992. doi: 10.3390/app12030992
[15] MILENKOVIĆ B, JOVANOVIĆ Đ, KRSTIĆ M. An application of Dingo Optimization Algorithm (DOA) for solving continuous engineering problems[J]. FME Transactions, 2022, 50(2): 331 − 338. doi: 10.5937/fme2201331M
[16] ALIKHAN J S, ALAGESWARAN R, AMALI S M J. Dingo optimization based network bandwidth selection to reduce processing time during data upload and access from cloud by user[J]. Telecommunication Systems, 2023, 83: 189 − 208.
[17] 王克鸿, 沈莹吉, 钱锋, 等. MAG焊熔池图像特征及可用信息分析[J]. 焊接学报, 2006, 27(11): 53 − 56. WANG Kehong, SHEN Yingji, QIAN Feng, et al. Characterization of MAG weld pool images and analysis of available information[J]. Transactions of the China Welding Institution, 2006, 27(11): 53 − 56.
[18] 冯志强, 袁浩, 刘鹏, 等. 变间隙MAG焊熔透状态的粗糙-模糊控制方法[J]. 焊接学报, 2023, 44(11): 22 − 35. FENG Zhiqiang, YUAN Hao, LIU Peng, et al. A rough-fuzzy control method for the penetration state of variable gap MAG welding[J]. Transactions of the China Welding Institution, 2023, 44(11): 22 − 35.
[19] KRIZHEVSKY A, SUTSKEVER I, HINTON G E. Image net classification with deep convolutional neural networks [C]//Advances in Neural Information Processing Systems, 2012: 1079 − 1105.
[20] SIMONYAN K, ZISSERMAN A. Very deep convolutional networks for large-scale image recognition[J]. arXiv preprint arXiv, 2014, 1409.1556.
[21] JOSE M, KUMAR S S, SHARMA A. Vibration assisted welding processes and their influence on quality of welds[J]. Science and Technology of Welding and Joining, 2016, 21(4): 243 − 258. doi: 10.1179/1362171815Y.0000000088
计量
- 文章访问数: 142
- HTML全文浏览量: 16
- PDF下载量: 30
