Research on dust identification and concentration detection method based on machine vision

Aiming at the problem that the current machine vision algorithm fails to combine position information with concentration value in the field of dust detection, we propose an algorithm that combines improved YOLOv5 with multivariate model. Firstly, a set of simulation experiment platform for collectin...

Full description

Saved in:
Bibliographic Details
Main Authors: Luyang TU, Qinghua CHEN, Yingsong CHENG, Bingyou JIANG
Format: Article
Language:zho
Published: Editorial Office of Safety in Coal Mines 2025-08-01
Series:Meikuang Anquan
Subjects:
Online Access:https://www.mkaqzz.com/cn/article/doi/10.13347/j.cnki.mkaq.20240469
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Aiming at the problem that the current machine vision algorithm fails to combine position information with concentration value in the field of dust detection, we propose an algorithm that combines improved YOLOv5 with multivariate model. Firstly, a set of simulation experiment platform for collecting and making dust data set is built, and the dust position information data set and dust concentration data set are made respectively. Then, the improved YOLOv5 is used to train the dust position information data set to obtain the training weight. At the same time, the dust image in the dust concentration data set is transformed into different color spaces, and the color features and texture features are extracted. The relationship between these features and dust concentration is analyzed to establish a multivariate model. Finally, the multivariate model is combined with YOLOv5 to obtain the ability to identify dust location information and detect its concentration in real time. The experimental results show that the improved YOLOv5 dust recognition model improves the accuracy and recall rate by 2.6 % and 3.1 % respectively compared with the original model. The accuracy rate reaches 91.8 % and the recall rate reaches 90.8 %. After combining with the multivariate model, the algorithm obtains the dust concentration detection ability, the detection accuracy reaches 98.79 %, and the adjustment accuracy reaches 96.03 %.
ISSN:1003-496X