【作者】 万文骏； 郭强； 荆丰伟； 凌智；

【Author】 Wan Wenjun;Guo Qiang;Jing Fengwei;Ling Zhi;National Engineering Research Center of Advanced Rolling Technology of USTB;

【机构】 北京科技大学高效轧制国家工程研究中心；

【摘要】 屈服强度、抗拉强度、延伸率等带钢力学性能指标是带钢质量的重要指标。带钢力学性能的预测一方面能够减少手动检验力学性能的工作量,另一方面是实现带钢力学性能在线调整的前提条件,所以具有重大意义。本文初步探究了使用Weka平台进行数据挖掘实现带钢力学性能预测的可行性。文中以某热连轧生产线的SPA-H钢种的实测数据为样本,使用了C4.5决策树算法的分类功能实现数据挖掘,最终得到决策树形式的性能预测模型,对样本数据的准确分类率达到79.8%。更多还原

【Abstract】 Yield strength,tensile strength,elongation are important indicators of strip’s quality.On one hand,the mechanical properties prediction of strip can reduce the workload of manual testing;on the other hand,it is the precondition to achieve the online adjustment of mechanical properties of the steel strip. Therefore, realizing the offline mechanical properties prediction of strip is of great significance. This paper did some preliminary study on the feasibility of using Weka to realize offline mechanical properties prediction of the steel strip. In this paper, the real-time data set of SPA-H is from a hot strip rolling production line, and the decision tree algorithm C4.5 is used for data mining, finally a decision tree model is generated to realize offline mechanical properties prediction of striped steel. The model’s correct classification rate is 79.8144%.更多还原