基于LIBS技术的高炉生铁硅含量快速检测系统研究

In this project, a detection system based on laser-induced breakdown spectroscopy (LIBS) is established to detect the silicon content in pig iron sample. With the deep knowledge of characteristics of element emission spectrum, pig iron solidification and spectrum transmission, the system is developed on the basis of blast furnace ironmaking theory, solidification theory, spectrum technology, plasma physics, modern detection technology, computer vision technology and data compression processing technology. The hardware is designed under the unique ideas, which is accurate, safe, reliable and longevous. In the quantitative software design, it is firstly considered the influence of silicon segregation, surface morphology, hot metal temperature and blast furnace operation on silicon content in pig iron sample. The study of relationship between silicon segregation, surface morphology, hot metal temperature and silicon content is not only helpful to make the detection system more practical and more accurate, but also enriches the metallurgy theory and detection technology. The successful development of this system has important significance to shed new light on the mechanism of LIBS, metal solidification theory and physical chemistry in metallurgy. It will also promote the rapid detection equipment, which has independent intellectual property rights in China. The research results will also enrich the theory and practice of BF metallurgy, plasma physics, spectrochemistry and detection technology in China. This project will play an important role for improving the level of automation, intelligence and information for blast furnace in China and energy conservation and emission reduction.

本项目基于对激光诱导击穿光谱技术（LIBS）检测高炉生铁样品硅含量过程中元素发射光谱、生铁样品凝固、光谱传输等特点的深刻认识，结合高炉炼铁理论、钢铁凝固理论、光谱技术、等离子体物理、现代检测技术、计算机视觉技术及数据压缩处理技术等，采用独特设计思想开发精准安全、可靠、长寿的硬件，在定量化软件设计中首次考虑了凝固过程硅元素偏析、样品表面形貌、铁水温度及高炉操作过程对生铁样品硅含量的影响，研发基于LIBS技术的高炉生铁硅含量快速检测系统。以上研究不仅使该系统更实用、检测更精准，并丰富了冶金理论及检测技术。生铁样品硅含量快速检测系统的成功研发将进一步揭示激光诱导击穿光谱技术、金属凝固及冶金物理化学的机理，对形成具有我国自主知识产权快速检测设备有重要价值。研究成果将丰富冶金理论、等离子体物理、光谱化学及检测技术，对提高高炉自动化、智能化、信息化水平及节能减排有重要经济和社会效益。

（1）项目研究内容如期完成，达到了预期目标，发表SCI检索文章4篇，EI检索文章4篇，申请受理发明专利2项，已授予实用新型专利1项，已批准软件著作权1项，在申请软件著作权1项。.（2）从热力学和动力学的角度对高炉中下部硅元素的传输机理进行了详细分析，基于冶金原理对高炉生产过程中影响硅含量的因素进行了筛选，为后续建立硅含量预测模型奠定了基础，既保证了自变量与因变量之间的高度相关性，又减少了计算量。.（3）考虑到高炉各参数与Si含量的滞后性以及高炉本身的动态特性，建立了基于遗传算法-在线核极限学习机的在线预测模型，该模型能够跟踪高炉的变化，从而对模型参数进行动态调整，为高炉操作者及时调整作业参数稳定炉况提供了指导。.（4）搭建基于LIBS技术的生铁样品快速测硅系统，优化了系统参数，提出使用PCA-ANN算法，使用多条谱线参与定量分析建模，降低了基体效应带来的影响。PCA-ANN方法能够在降低数据维度，减少计算量的基础上对输入变量和Si含量之间的非线性关系进行建模，取得了较好的定标效果和预测效果。.（5）将LIBS技术应用于铁水样品中Si、Mn、Ti元素的偏析情况同步检测，证实了LIBS技术今后在大型金属产品元素偏析检测领域的可行性。.（6）编写了高炉铁水硅含量在线预测软件和LIBS系统控制软件，为以上技术早日大规模应用于高炉生产奠定了基础。.本项目建立的高炉铁水硅含量在线预测模型和快速检测系统有利于高炉现场操作人员及时了解高炉炉缸热状态，将铁水硅含量稳定在较低水平，实现高产、低耗、低成本的目标。