[1]张 豪,郑 迪,王陈佳倍.溯源鹰眼抑尘算法在智能环保管控平台中应用[J].工业仪表与自动化装置,2025,(06):95-101.[doi:10.19950/j.cnki.CN61-1121/TH.2025.06.017]
　ZHANG Hao,ZHENG Di,WANG Chenjiabei.The traceable eagle-eye dust suppression algorithm using in intelligent environmental protection and control platforms[J].Industrial Instrumentation & Automation,2025,(06):95-101.[doi:10.19950/j.cnki.CN61-1121/TH.2025.06.017]
点击复制

溯源鹰眼抑尘算法在智能环保管控平台中应用()

《工业仪表与自动化装置》[ISSN:1000-0682/CN:61-1121/TH]

卷:
期数:
2025年06期
页码:
95-101
栏目:
出版日期:
2025-12-15

文章信息/Info

Title:
The traceable eagle-eye dust suppression algorithm using in intelligent environmental protection and control platforms
文章编号:
1000-0682（2025）06-0095-07
作者:
张 豪; 郑 迪; 王陈佳倍
中冶南方都市环保工程技术股份有限公司，湖北 武汉 430205
Author(s):
ZHANG Hao;  ZHENG Di;  WANG Chenjiabei
（CCEPC, Hubei Wuhan 430205,China）
关键词:
环保管控; 鹰眼抑尘; 目标检测; 动作检测; 云边协同部署
Keywords:
environmental management;  eagle-eye dust suppression;  object detection;  action recognition;  cloud-edge collaborative deployment
分类号:
X513
DOI:
10.19950/j.cnki.CN61-1121/TH.2025.06.017
文献标志码:
A
摘要:
针对料场扬尘智能管控中现有鹰眼抑尘算法独立部署导致的识别准确率低和维护升级困难等问题，该文创新性地提出一种融合目标检测和动作识别的溯源鹰眼抑尘算法，通过Docker容器与边缘计算实现算法的云边协同灵活部署。具体而言，开发了基于微服务架构的算法管理模块，实现算法镜像集中管理与远程快速升级，边缘端容器化快速部署算法并实时响应现场扬尘情况。测试结果显示，目标检测准确率白天约99.31%、夜晚86.67%，动作识别准确率白天约86.71%、夜晚92.59%，整体准确率达到87.65%。相比现有研究普遍采用的单一目标检测算法（准确率约79%-80%），该方法的整体识别准确率提高了约7%-10%。该方案在性能和部署效率上均表现出明显优势，为料场超低排放改造提供了高效、经济且可持续的技术路径。
Abstract:
Addressing the issues of low accuracy and difficult maintenance of existing Eagle-Eye dust suppression algorithms deployed independently in intelligent dust management systems, this paper innovatively proposes a traceable Eagle-Eye dust suppression algorithm integrating object detection and action recognition, achieving flexible cloud-edge collaborative deployment through Docker containers and edge computing. Specifically, an algorithm management module based on a microservice architecture was developed, allowing centralized management and rapid remote upgrading of algorithm images, while enabling edge-side containerized rapid deployment and real-time response to dust conditions onsite. Test results showed daytime object detection accuracy of approximately 99.31% and nighttime 86.67%, while action recognition accuracy reached about 86.71% during the day and 92.59% at night, with an overall accuracy of 87.65%. Compared with the commonly used single-object detection algorithms (accuracy around 79%-80%), the proposed method improved overall accuracy by approximately 7%-10%. The proposed approach demonstrates clear advantages in performance and deployment efficiency, providing an effective, economical, and sustainable technical solution for ultra-low emission transformation in material yards.

参考文献/References:

[1] 生态环境部等.关于推进实施钢铁行业超低排放的意见[R].环大气[2019]35号,2019.

[2] 中国环保产业协会.钢铁企业超低排放改造技术指南[R].中环协[2020]4号,2020.
[3] 蔡恒忠,黄玉鸿,柯雄,韩明爽.钢铁行业超低排放管控平台建设浅析[J].环境工程,2023,41(S1): 535-538+569.
[4] 王珲,安登飞,张六零.钢铁企业烟粉尘重点排放源分析[J].环境工程,2016,34(3): 62-64+101.
[5] 李森萍,冯建行.基于多源数据融合的建筑施工扬尘污染特征识别研究[J].环境科学与管理,2022,47(4): 59-63.
[6] 郑晓宇,徐宇锋,王倩.企业智慧环保管控平台的设计与构建[J].中国科技投资,2023(1): 31-33.
[7] 陈三清,张凯迪,王盛勇,等.基于企业网格化管理的智慧环保管控平台建设研究[J].工业安全与环保,2023,49(9): 100-102.
[8] 李毅仁,邢奕,孙宇佳,等.钢铁工业低碳绿色发展路径与实践[J].工程科学学报,2023,45(9): 1583-1591.
[9] 甄然,刘颖,孟凡华,等.基于YOLOv5的轻量化目标检测算法[J].无线电工程,2023,53(6): 1242-1250.
10] 黄同愿,谭禹,朱金江.基于SlowFast网络的孤立词手语识别算法研究.重庆理工大学学报(自然科学),2023(12): 267-275.
[11] 单玉忠,董为,韩梦豪,等.工业互联网平台边云协同系统建设与应用[J].计算机应用,2022,42(S2): 135-139.
[12] 戴文斌.基于微服务与容器化的工业边缘计算应用设计[J].信息与控制,2021,38(2): 30-32.
[13] 邬贺铨.边缘计算助力工业互联网[J].信息与控制,2023,40(2): 10-13.
[14] 王艳,张游杰.基于改进残差网络的扬尘图像识别方法[J].计算机系统应用,2021,30(5): 202-207.

相似文献/References:

备注/Memo

备注/Memo:
收稿日期：2025-05-27第一作者：张豪(1986 — )，男，湖北武汉人，硕士，高级工程师， 研究方向为智慧发电和智能环保管控系统。
更新日期/Last Update: 1900-01-01