基于数学模型分析的带钢生产工艺优化与质量控制研究  

Research on Optimization of Steel Strip Production Process and Quality Control Based on Mathematical Modeling

作　　者：刘冰 邱旭 丁玉民 孙昊 

机构地区：[1]辽宁科技学院冶金与材料工程学院,辽宁本溪117004

出　　处：《辽宁科技学院学报》2024年第5期13-17,共5页Journal of Liaoning Institute of Science and Technology

基　　金：辽宁省教育厅面上项目“KX84系列牙爪大跑道接触疲劳分析与预防”(LJKZ1067);技术开发项目“轴承钢表面性能研究”。

摘　　要：本研究以钢铁产品质量优化为背景,聚焦冷轧带钢连续退火工序。因工艺参数耦合难以建立机理模型,故通过数据分析,确定影响带钢机械性能的参数,以带钢硬度为指标分析参数相关性,建立随机森林模型和神经网络模型进行带钢产品质量在线检测,对比性能选优。利用建立的模型迭代已知参数范围,解码未知参数的硬度值以优化工艺参数。该研究建立数据驱动模型,为钢铁企业提供有效方案,可提高产品质量、降低成本、提升效率,从而促进钢铁行业发展。In the background of optimizing steel product quality,the paper was focused on the continuous annealing process of cold-rolled steel strip.Due to the complex coupling of process parameters and thus difficulties in establishing a mechanistic model,data analysis was conducted to identify parameters affecting the mechanical properties of steel strip.Using the hardness of steel strip as an indicator,the correlations between parameters were analyzed.Random Forest and neural network models were established for online quality detection,with performance comparisons made to select the optimal model.The established models were used to iterate over known parameter ranges and to decode the hardness values of unknown parameters in order to optimize the process parameters.The current paper establishes data-driven models providing effective solutions for steel companies,which can improve product quality,reduce costs,enhance efficiency,and promote the development of the steel industry.

关 键 词：随机森林模型 神经网络模型 机械性能 相关性 带钢 

分 类 号：TG156.2[金属学及工艺—热处理]