基于交叉注意力的多源数据融合的气体泄漏检测

doi:10.16265/j.cnki.issn1003-3033.2024.07.0135

摘要/Abstract

摘要：

为解决单一传感器在管道气体泄漏检测时容易出现误报、漏报的问题,及时预警并反馈泄漏状况,提出一种基于交叉注意力的多源数据融合管道泄漏检测方法。首先,利用预训练的ShuffleNetV2模型提取热像仪数据的空间特征;然后,结合一维卷积神经网络(1DCNN)和双向门控循环单元(BiGRU),构建1DCNN-BiGRU模型,以提取气体传感器数据的时序特征;最后,运用交叉注意力捕获数据的时空关联性得到2个数据源的特征表示,通过残差方式进行特征连接后输入到分类层中,得到识别结果。结果表明:所构建的多源数据融合模型(SCGA)对气体识别准确率为99.22%,损失值在0~0.04内波动;与仅使用气体传感器数据的支持向量机(SVM)、1DCNN、BiGRU模型相比,准确率至少提升4.12%;与仅使用热图像传感器数据的MobileNetV3、ShuffleNetV2、ResNet18模型相比,准确率至少提升1.14%;与将时序特征和空间特征直接拼接的多源数据融合模型(SCG)相比,准确率提升1%。SCGA模型对气体识别具有较高精度。

关键词: 交叉注意力, 多源数据融合, 气体泄漏检测, 卷积神经网络(CNN), 双向门控循环单元(BiGRU)

Abstract:

In order to solve the problem of false alarms and missed alarms in pipeline gas leakage detection using a single sensor, timely warning and feedback of leakage status, a multi-source data fusion pipeline leakage detection method based on cross-attention was proposed. Firstly, the pre-trained ShuffleNetV2 model was used to extract spatial features from thermal imaging data. Then, a 1DCNN BiGRU model was constructed by combining a one-dimensional CNN (1DCNN) and a bidirectional gated recurrent unit (BiGRU) to extract temporal features from gas sensors. Finally, cross-attention was used to capture the spatiotemporal correlation of the data and obtain the feature representations of the two data sources. The residual method was used to connect the features and input them into the classification layer to obtain the recognition results. The results show that the constructed SCGA model has a gas recognition accuracy of 99.22%, and the loss value fluctuates between 0-0.04. Compared with support vector machines (SVM), 1DCNN, and BiGRU models that only use gas sensor data, the accuracy is improved by at least 4.12%. Compared with MobileNetV3, ShuffleNetV2, and ResNet18 models that only use thermal image sensor data, the accuracy is improved by at least 1.14%. Compared with the multi-source data fusion model SCG, which simply connects temporal and spatial features, the accuracy is improved by 1%. It was verified that the SCGA model has high accuracy.

Key words: cross-attention, multi source data fusion, gas leak detection, convolutional neural network (CNN), bidirectional gated recurrent unit (BiGRU)

中图分类号:

X937

王新颖, 杨阳, 田豪杰, 陈俨, 张敏. 基于交叉注意力的多源数据融合的气体泄漏检测[J]. 中国安全科学学报, 2024, 34(7): 91-97.

WANG Xinying, YANG Yang, TIAN Haojie, CHEN Yan, ZHANG Min. Gas leak detection based on cross-attention multi-source data fusion[J]. China Safety Science Journal, 2024, 34(7): 91-97.

图/表 8

图1

图2

表1

图3

图4

表2

表3

图5

参考文献 20

[1]	中华人民共和国中央人民政府. 中华人民共和国国民经济和社会发展第十四个五年规划和2035年远景目标纲要[EB/OL]. (2021-03-13). http://www.gov.cn/xinwen/2021-03/13/content_5592681.htm.
[2]	向鹏成, 张子薇, 李卉, 等. 城市地下燃气管道泄漏事故致因因素系统分析[J]. 中国安全科学学报, 2023, 33(12):140-147. doi: 10.16265/j.cnki.issn1003-3033.2023.12.2162
	XIANG Pengcheng, ZHANG Ziwei, LI Hui, et al. Systematic analysis on causative factors for gas pipeline leakage accidents in urban underground[J]. China Safety Science Journal, 2023, 33(12):140-147. doi: 10.16265/j.cnki.issn1003-3033.2023.12.2162
[3]	ATTALLAH O, MORISI I. An electronic nose for identifying multiple combustible/harmful gases and their concentration levels via artificial intelligence[J]. Measurement, 2022,199:DOI:10.1016/j.measurement.2022.111458.
[4]	李龙, 应滨州, 李华曜, 等. 电子鼻多组分气体检测算法与性能研究[J]. 仪表技术与传感器, 2023(11):1-6,11.
	LI Long, YING Binzhou, LI Huayao, et al. Algorithm and performance study of electronic nose for multi-component gas detection[J]. Instrument Technique and Sensor, 2023(11):1-6,11.
[5]	RAHMAN C A, ROGERS S, et al. Tensor-based approach for liquefied natural gas leakage detection from surveillance thermal cameras: a feasibility study in rural areas[J]. IEEE Transactions on Industrial Informatics, 2021, 17(12): 8122-8130.
[6]	WANG Jingfan, JI Jingwei, RAVIKUMAR A P, et al. VideoGasNet: deep learning for natural gas methane leak classification using an infrared camera[J]. Energy, 2022, 238: DOI:10.1016/j.energy.2021.121516.
[7]	NARKHEDE P, WALAMBE R, MANDAOKAR S, et al. Gas detection and identification using multimodal artificial intelligence based sensor fusion[J]. Applied System Innovation, 2021, 4(1):DOI:10.3390/asi4010003.
[8]	杨余, 杨鑫, 王英, 等. 基于mini-1D-CNN模型的TE过程故障诊断[J]. 中国安全科学学报, 2023, 33(2):173-178. doi: 10.16265/j.cnki.issn1003-3033.2023.02.0017
	YANG Yu, YANG Xin, WANG Ying, et al. Fault diagnosis of TE process based on mini-1D-CNN model[J]. China Safety Science Journal, 2023, 33(2):173-178. doi: 10.16265/j.cnki.issn1003-3033.2023.02.0017
[9]	梁宏涛, 刘硕, 杜军威, 等. 深度学习应用于时序预测研究综述[J]. 计算机科学与探索, 2023, 17(6):1285-1300. doi: 10.3778/j.issn.1673-9418.2211108
	LIANG Hongtao, LIU Shuo, DU Junwei. et al. A review of the application of deep learning in time series prediction research[J]. Journal of Frontiers of Computer Science and Technology, 2023, 17(6):1285-1300.
[10]	HAN Jin, YANG Yonghao. L-Net: lightweight and fast object detector-based ShuffleNetV2[J]. Journal of Real-time Image Processing, 2021, 18(6): 2527-2538.
[11]	JI Jundong, XU Cheng, ZHANG Xiaodan, et al. Spatio-temporal memory attention for image captioning[J]. IEEE Transactions on Image Processing, 2020, 29: 7615-7628.
[12]	田宝柱, 徐文涛, 梁鹏, 等. 基于多传感器融合的埋地输水管道泄漏声发射定位方法[J]. 科学技术与工程, 2023, 23(24):10307-10 316.
	TIAN Baozhu, XU Wentao, LIANG Peng, et al. Acoustic emission localization method of water pipeline based on multi-sensor fusion[J]. Science Technology and Engineering, 2023, 23(24):10307-10 316.
[13]	王欣, 干镞锐, 许雅玺, 等. 基于字词向量融合的民航智慧监管短文本分类[J]. 中国安全科学学报, 2024, 34(2):37-44. doi: 10.16265/j.cnki.issn1003-3033.2024.02.0121
	WANG Xin, GAN Zurui, XU Yaxi, et al. Short text classification of civil aviation intelligent supervision based on character-word fusion[J]. China Safety Science Journal, 2024, 34(2):37-44. doi: 10.16265/j.cnki.issn1003-3033.2024.02.0121
[14]	赵宏, 郭岚, 陈志文, 等. 基于多模态融合与多层注意力的视频内容文本表述研究[J]. 计算机工程, 2022, 48(10):45-54. doi: 10.19678/j.issn.1000-3428.0063294
	ZHAO Hong, GUO Lan, CHEN Zhiwen, et al. Research on video content text representation based on multimodal fusion and multilayer attention[J]. Computer Engineering, 2022, 48(10):45-54. doi: 10.19678/j.issn.1000-3428.0063294
[15]	NARKHEDE P, WALAMBE R, CHANDEL P, et al. MultimodalGasData: multimodal dataset for gas detection and classification[J]. Data, 2022, 7(8):DOI:10.3390/data7080112.
[16]	朱馨, 李建微, 郭伟, 等. 基于机器学习的森林火险预测模型[J]. 中国安全科学学报, 2022, 32(9):152-157. doi: 10.16265/j.cnki.issn1003-3033.2022.09.2729
	ZHU Xin, LI Jianwei, GUO Wei, et al. Forest fire risk prediction model based on machine learning[J]. China Safety Science Journal, 2022, 32(9):152-157. doi: 10.16265/j.cnki.issn1003-3033.2022.09.2729
[17]	梁秀满, 安金铭, 曹晓华, 等. 基于改进MobileNetV3烧结断面火焰图像识别[J]. 电子测量技术, 2023, 46(14):182-187.
	LIANG Xiuman, AN Jinming, CAO Xiaohua, et al. Flame image recognition of sintering section based on improve MobileNetV3[J]. Electronic Measurement Technology, 2023, 46(14):182-187.
[18]	刘飞, 陈仁文, 邢凯玲, 等. 基于迁移学习与深度残差网络的滚动轴承快速故障诊断算法[J]. 振动与冲击, 2022, 41(3):154-164.
	LIU Fei, CHEN Renwen, XING Kailing, et al. Fast fault diagnosis algorithm for rolling bearing based on transfer learning and deep residual networks[J]. Journal of Vibration and Shock, 2022, 41(3):154-164.
[19]	RAHATE A, MANDAOKAR S, CHANDEL P, et al. Employing multimodal co-learning to evaluate the robustness of sensor fusion for industry 5.0 tasks[J]. Soft Computing, 2023, 27(7): 4139-4155.
[20]	ATTALLAH O. Multitask deep learning-based pipeline for gas leakage detection via E-nose and thermal imaging multimodal fusion[J]. Chemosensors, 2023, 11(7):DOI:10.3390/chemosensors11070364.

模型	参数	取值
CNN	CNN-1出入通道 CNN-2出入通道卷积核填充	[1,16] [16,32] 2 1
池化层	卷积核填充	2 1
BiGRU	输入通道隐藏维度	32 64
交叉注意力	头数量融合特征维度	4 128
网络训练	学习率批次大小优化器迭代次数损失函数	0.001 32 Adam 50 Cross Entropy Loss

文章	方法	准确率/%
文献[7]	CNN+LSTM 早期融合	96.0
文献[19]	CNN+LSTM 中期融合	94.5
文献[19]	CNN+LSTM 多任务融合	96.9
文献[20]	Inception+DWT+ BiLSTM 中期融合	98.5
文献[20]	(ResNet50+Inception+ MobileNet)+DWT+ DCT+BiLSTM 多任务融合	99.2
本文	ShuffleNet+ (CNN-BiGRU)+CA	99.2