基于卷积神经网络的12Cr1MoV钢金相组织球化级别智能化分析
Intelligent Analysis of 12Cr1MoV Steel Metallographic Structure Sphericity Level Based on Convolutional Neural Network
赵隆 1韩小稚 1王景人 1寇威1
作者信息
- 1. 陕西省特种设备检验检测研究院 西安 710048
- 折叠
摘要
为了高效准确地评定火电厂 12Cr1MoV钢金相组织的球化级别,先对原始金相组织图像进行灰度值归一化和图像去噪等处理,建立金相组织图像数据库,再采用 4 种不同的卷积神经网络模型对金相组织图像进行分类识别,开发了金相组织智能分析软件.实验结果表明:Inception-v3 模型识别 12Cr1MoV钢金相组织的球化级别的准确度可达到 93%.开发的智能分析软件可以自动、准确和高效地评定球化级别,为火电厂锅炉材料检验提供了更可靠的金相分析工具.
Abstract
In order to efficiently and accurately evaluate the sphericity level of 12Cr1MoV steel in thermal power plant,the original metallograph tissue images were normalized and denoised,and the metallographic tissue image database was established,and then four different convolutional neural network models were used to classify and identify the metalgraphic tissue image,and the metallographic tissue intelligent analysis software was developed.The experimental results show that the Inception-v3 model can identify the sphericity level of the metallographic organization of 12Cr1MoV steel with an accuracy of 93%.The intelligent analysis software can assess the sphericity level automatically,accurately and efficiently,providing a more reliable metallographic analysis tool for boiler material inspection in thermal power plant.
关键词
金相组织/深度学习/卷积神经网络/图像处理Key words
Metallographic organization/Deep learning/Convolutional neural network/Image processing引用本文复制引用
出版年
2024
NETL
NSTL
万方数据