摘要
针对复杂工况下煤矸识别效率低、分选难度大的问题,采用VGG-16网络搭建煤矸识别模型,对煤矸识别模型的识别准确率和识别环境影响因素进行了研究,并对VGG-16煤矸识别模型进行了优化。结果表明:(1)优化后的VGG-16网络模型准确率为97.00%,单张煤矸图像识别时间为0.0697s,单张煤矸图像识别所用时间缩短了0.85%;(2)在不同水分、灰分和粉尘等环境因素下,煤矸识别模型的准确率均达到95%以上,其中水分对模型的识别准确率影响最大,表面浸润30 s比干燥的识别准确率低2.01个百分点;(3)鉴于煤与矸石的共伴生特性,对煤表面夹矸、矸表面带煤两种复杂情况进行了煤矸有效识别。研究表明:优化后的VGG-16网络模型具有一定的抗干扰能力,可以实现复杂情况下煤矸的高效精准识别,可为后续煤矸石智能化分选提供理论基础和技术支撑。
Aiming at the problems of low identification efficiency and difficulty in separating coal gangue under complex working conditions,the VGG-16 network was used to build a coal gangue identification model.The identification accuracy of the coal gangue identification model and the identification of environmental factors were studied,and the VGG-16 coal gangue identification model was optimized.The results are show as follows.Firstly,the accuracy of the optimized VGG-16 network model is 97.00%,the identification time of a single coal gangue image is 0.0697 s,and the identification time of a single coal gangue image is shortened by 0.85%.Secondly,under different environmental factors such as moisture,ash and dust,the accuracy of the coal gangue identification model is more than 95%.Among them,moisture has the greatest influence on the identification accuracy of the model.The identification accuracy of surface wetting for 30 s is 2.01 percentage points lower than that of drying.Finally,in view of the co-associated characteristics of coal and gangue,the effective identification of coal gangue was carried out for two complex situations of coal surface with gangue and gangue surface with coal.The research shows that the optimized VGG-16 network model has certain anti-interference ability,which can realize the efficient and accurate identification of coal gangue under complex conditions,and can provide theoretical basis and technical support for the subsequent intelligent separation of coal and gangue.
作者
黄可
樊玉萍
董宪姝
马晓敏
HUANG Ke;FAN Yuping;DONG Xianshu;MA Xiaomin(College of Mining Engineering,Taiyuan University of Technology,Taiyuan,Shanxi 030024,China;State Key Laboratory of Mineral Processing,Beijing 100160,China)
出处
《矿业研究与开发》
CAS
北大核心
2024年第9期219-226,共8页
Mining Research and Development
基金
国家自然科学基金国际(地区)合作与交流项目(51820105006)
山西省重点研发计划项目(202202090301009)。
关键词
煤矸识别
VGG-16网络模型
识别准确率
环境因素
Coal gangue identification
VGG-16 network model
Identification accuracy
Environmental factor
