基于CNN-GRU模型的烟气黑度监测研究  

Research on Smoke Blackness Detection Based on the CNN-GRU Model

作　　者：党祥斌 李楠 孙泽龙 敬晨曦 Dang Xingbin;Li Nan;Sun Zelong;Jing Chenxi(Ordnance Industry Hygiene Research Institute,Xi’an 710000,China;Xi’an Technological University,School of Weapon Science and Technology,Xi’an 710021,China)

机构地区：[1]兵器工业卫生研究所,陕西西安710000 [2]西安工业大学兵器科学与技术学院,陕西西安710021

出　　处：《皮革制作与环保科技》2024年第19期124-126,共3页Leather Manufacture and Environmental Technology

摘　　要：为了有效、准确地监测烟气黑度,本研究提出了一种基于卷积神经网络(CNN)和门控循环单元(GRU)的烟气黑度监测方法。该方法结合了CNN在图像处理方面的优势和GRU在时序数据处理方面的能力,能够有效地从烟气图像中提取特征并实现实时监测。首先,对烟气图像进行预处理,包括去噪、增强和转换等步骤,以提高图像质量。然后,利用CNN对预处理后的图像进行特征提取,并通过GRU对时序特征进行建模。最后,通过构建分类器对黑度等级进行预测。模型结果表明,该方法在烟气黑度监测方面具有较高的准确性和实时性,为环保监测和工业生产提供了有效支持。To eff ectively and accurately monitor smoke blackness,a monitoring method based on Convolutional Neural Networks(CNN)and Gated Recurrent Units(GRU)is proposed.This method combines the advantages of CNN in image processing with the capabilities of GRU in sequential data handling,enabling eff ective feature extraction from smoke images and real-time monitoring.Initially,smoke images undergo preprocessing,including noise reduction,enhancement,and transformation,to improve image quality.Subsequently,the preprocessed images are subject to feature extraction through CNN,followed by sequential feature modeling using GRU.Finally,a classifi er is constructed to predict the blackness level.The model results demonstrate that this method achieves high accuracy and realtime performance in smoke blackness monitoring,providing eff ective support for environmental monitoring and industrial production.

关 键 词：深度学习 模型 烟气黑度 

分 类 号：TS5[轻工技术与工程—皮革化学与工程]