基于卷积神经网络的客运索道承载索断丝损伤定量识别方法研究
Research on Quantitative Identification Method of Broken Wire Damage of Passenger Ropeway Load Carrying Rope Based on Convolutional Neural Network
万强 1孙润业 1杨志学 2张铁骥 2王宝轩1
作者信息
- 1. 中国特种设备检测研究院 北京 100029
- 2. 吉林省特种设备检验研究院 长春 130103
- 折叠
摘要
为了提高客运索道承载索断丝损伤定量识别精度,本文研究提出了一种基于卷积神经网络的断丝损伤定量识别方法.通过对卷积神经网络定量识别模型结构和训练参数进行选取和优化,将得到的目标模型对不同钢丝绳尺寸、断丝数目、断口宽度、内外部的断丝损伤进行定量识别,实验结果表明该模型能准确识别各类断丝损伤,分类准确率达到 99%以上.最后通过现场应用进一步证实了所提方法的有效性、准确性和适用性.
Abstract
To improve the precision of quantifying the damage caused by broken wires of passenger ropeways load carrying rope,this paper presents a novel approach for the quantitative identification of such damage,utilizing convolutional neural networks.Through the selection and optimization of the structure and training parameters of the quantitative identification model of convolutional neural network,the obtained target model was tested for quantitative identification of broken wire damage with different diameters,different number of broken wires,different break widths and internal and external.The results show that the model can accurately identify various types of broken wire damage,and the classification accuracy is more than 99%.Finally,the validity,accuracy and applicability of the proposed method are further verified by field application.
关键词
卷积神经网络/客运索道/断丝损伤/定量识别Key words
Convolutional neural network/Passenger ropeway/Broken wire damage/Quantitative identification引用本文复制引用
出版年
2024
NETL
NSTL
万方数据