[1]李俊芹,景世沛.doi: 10.3969/j.issn.1001-3849.2026.05.015基于机器视觉的电镀污泥成分预测模型构建[J].电镀与精饰,2026,(05):105-103.
　LI Junqin,JING Shipei.Fabrication of prediction model for electroplating sludge composition based on machine vision[J].Plating & Finishing,2026,(05):105-103.
点击复制

doi: 10.3969/j.issn.1001-3849.2026.05.015基于机器视觉的电镀污泥成分预测模型构建()

《电镀与精饰》[ISSN:1001-3849/CN:12-1096/TG]

卷:
期数:
2026年05
页码:
105-103
栏目:
出版日期:
2026-05-31

文章信息/Info

Title:
Fabrication of prediction model for electroplating sludge composition based on machine vision
作者:
李俊芹1; 景世沛2
(1. 郑州电子信息职业技术学院,信息工程学院,河南 郑州 451450 ;2. 河南牧业经济学院,信息工程学院,河南 郑州 450046)
Author(s):
LI Junqin1;  JING Shipei2
(1. School of Information Engineering, Zhengzhou Professional Technical Institute of Electronic & Information, Zhengzhou 451450, China; 2. School of Information Engineering, Henan University of Animal Husbandry and Economy, Zhengzhou 450046, China)
关键词:
机器视觉; 随机森林(RF); 多元线性回归(MLR); 重金属预测
Keywords:
machine vision;  random forest (RF) ;  multiple linear regression (MLR) ;  heavy metal prediction
分类号:
TQ153.2；TG17；TP181
文献标志码:
A
摘要:
电镀污泥作为电镀行业固体废料，含Cr、Fe等多种重金属，传统上依赖实验室的化学检测手段进行分析。因流程复杂、耗时较长，难以适配其资源化回收过程中的动态监测需求。为了提升电镀污泥重金属含量检测效率与精准度，提出了基于机器视觉的快速预测模型，通过采集多来源电镀污泥样品，经烘干、研磨、压片预处理后，借助能量色散型X射线荧光光谱法(EDXRF)获取光谱数据，经归一化、离散小波变换预处理提取特征峰净强度，分别建立多元线性回归(MLR)与随机森林(RF)模型进行分析。结果表明，基于实验的150条原始谱图，随机森林模型对Cr、Fe、Cu、Ni、Zn的分类准确率达100%，Fe元素的均方根误差为0.202%，预测Cr元素的决定系数为0.994；多元线性回归模型预测Cr元素的决定系数为0.954。其中，多元线性回归模型适用于基础快速分析场景，随机森林模型更适配复杂基体下的精准分析需求，研究构建的双模型体系为电镀污泥重金属检测提供新路径。
Abstract:
Electroplating sludge, a solid waste from the electroplating industry, contains various heavy metals such as Cr and Fe. Conventional chemical detection methods relying on laboratory analysis are difficult to adapt to the dynamic monitoring needs of the resource recovery process due to their complex procedures and time consuming nature. To improve the efficiency and accuracy of heavy metal content detection in electroplating sludge, it proposes a machine vision based on rapid prediction model. Multi source electroplating sludge samples were collected, dried, ground, and compressed into tablets for preprocessing. Energy dispersive X-ray fluorescence spectroscopy (EDXRF) was employed to obtain spectral data, which underwent normalization and discrete wavelet transform preprocessing to extract net characteristic peak intensities. Multiple linear regression (MLR) and random forest (RF) models were established accordingly. The results indicate that the RF model achieved a classification accuracy of 100% for Cr, Fe, Cu, Ni, and Zn based on 150 original spectra from experiments. The root mean square error for Fe is only 0.202, and the coefficient of determination for predicting Cr reaches 0.994. In comparison, the MLR model achieves a coefficient of determination of 0.954 for predicting Cr. The MLR model is suitable for basic rapid analysis scenarios, while the RF model is better adapted for precise analysis under complex matrix conditions. The dual model systems constructed in this study provide new approaches for heavy metal detection in electroplating sludge
更新日期/Last Update: 2026-05-12