摘要
永磁直驱带式输送机煤量的精确检测是节能调速的重要依据,针对不同环境下煤量的快速检测问题,结合迁移学习方法和密集连接思想提出一种基于VGG16改进的FF-CNN网络。将该网络作为煤量图像的特征提取器,实现煤量的检测分类,将煤量图像分为“无煤量”、“少煤量”、“中煤量”、“多煤量”和“满载量”。实验结果表明,训练后的FF-CNN网络检测准确率达98.2%,平均检测1张图像的时间为0.058 s,满足检测精度和时效性要求。
The accurate measurement of coal quantity of permanent magnet direct drive belt conveyor is an important basis for energy-saving and speed regulation. To solve the problem of rapid detection of coal quantity under different environments, a FF-CNN Network based on VGG16 is proposed, which combines transfer learning method and dense connection idea. It is used as the feature extractor of coal quantity image to realize the detection and classification of coal quantity, and the coal quantity image is divided into no coal quantity, small coal quantity, medium coal quantity, large coal quantity and full coal quantity. The experimental results show that the trained FF-CNN network detection accuracy reaches 98.2%, and the average detection time of an image is 0.058 s, which meets the requirements of detection accuracy and timeliness.
作者
王桂梅
李学晖
杨立洁
刘杰辉
WANG Guimei;LI Xuehui;YANG Lijie;LIU Jiehui(School of Mechanical and Equipment Engineering,Hebei University of Engineering,Handan 056038,China)
出处
《煤炭技术》
CAS
北大核心
2022年第1期188-190,共3页
Coal Technology
基金
河北省自然科学基金项目(E2019402436)
关键词
永磁直驱带式输送机
煤量检测
卷积神经网络
特征提取
图像分类
permanent magnet direct drive belt conveyor
coal quantity detection
convolution neural network
feature extraction
image classification
