结合YOLOv3与K-Means算法的循环水场自动监测设备方法研究  

Research on the Method of Automatic Monitoring Equipment for Circulating Water Field Combining YOLOv3 and K-Means Algorithm

作　　者：冯泳 张福桂 陈邦 FENG Yong;ZHANG Fugui;CHEN Bang(Maoming Zhonghe Guosong Water Treatment Technology Co.,Ltd.Postal Code of Maoming City,Guangdong Province 525000,China)

机构地区：[1]茂名众和国颂水处理技术有限公司,广东茂名525000

出　　处：《自动化与仪器仪表》2025年第3期107-111,共5页Automation & Instrumentation

摘　　要：为了提高水资源利用率,减少水体污染,提出了一种基于改进YOLOv3循环水场自动监测系统,结合YOLOv3与K-Means算法构建循环水场水面垃圾检测模型,对循环水场的水质进行实时监测并对水面垃圾进行自动检测与识别。首先,选择YOLOv3作为水场水面垃圾检测的基础模型,并为了进一步提高模型的检测速度与精度进行改进;然后将设计模型部署到循环水场自动监测系统中,并对系统整体框架、软件、硬件进行设计;最后对垃圾检测模型与自动监测系统进行实验测试。测试结果表明:设计的循环水场水面垃圾检测模型的平均F1维持在0.96左右,平均精准度维持在97%左右,mAP值达到了97.06%,与传统的目标检测模型相比,具有可行性和优越性,更加适用于循环水场水面垃圾自动检测与识别工作;设计系统可以实时监测并显示水中的PH值、电导率、药剂浓度以及钙硬度等参数,同时及时对水面垃圾进行检测并识别垃圾类别,具有可行性和有效性。In order to improve the utilization rate of water resources and reduce water pollution,this paper proposes an improved YOLOv3 automatic monitoring system for circulating water plants.Combining YOLOv3 and K-Means algorithm,a water surface garbage detection model for circulating water plants is constructed to monitor the water quality in real time and automatically detect and identify surface garbage.Firstly,YOLOv3 is chosen as the basic model for water surface garbage detection in the water field,and improvements are made to further improve the detection speed and accuracy of the model;Then deploy the design model to the automatic monitoring system of the circulating water plant,and design the overall framework,software,and hardware of the system;Finally,experimental testing was conducted on the garbage detection model and automatic monitoring system.The test results show that the average F1 of the water surface garbage detection model designed in this paper is maintained at around 0.96,the average accuracy is maintained at around 97%,and the mAP value reaches 97.06%.Compared with traditional object detection models,it is feasible and superior,and more suitable for automatic detection and recognition of water surface garbage in circulating water plants;The system designed in this article can monitor and display real-time parameters such as pH value,conductivity,drug concentration,and calcium hardness in water,while also detecting and identifying garbage types on the water surface in a timely manner,which is feasible and effective.

关 键 词：自动监测 YOLOv3 K-MEANS算法 循环水场 目标检测 

分 类 号：TP23[自动化与计算机技术—检测技术与自动化装置]