【摘要】 以某钢铁企业的热轧带钢生产实际数据作为分析对象,基于改进的随机森林算法分析工艺参数与产品质量间的隐含关系,进行影响产品质量关键工艺参数的特征提取,建立热轧带钢产品缺陷预测模型.实验结果表明,对非平衡数据集进行平衡处理可以提高样本预测精度;采用CART与C4. 5相结合的方法比单一方法可以进一步提升预测精度;同时根据特征的高相关与低相关特性,将互信息作为评价指标应用于特征选择,可以提升随机森林算法的分类效果.在以上三种改进策略下,热轧带钢缺陷的识别率得到明显提高.更多还原

【Abstract】 The process data of hot-rolled strips from an iron and steel enterprise w ere analyzed to find out the inherent relationship betw een process parameters and production quality by using an improved random forests algorithm. After critical features being extracted,a defect prediction model w as built. According to the experiment,balancing operation can improve the prediction accuracy of the imbalanced data sets. M eanw hile,the combination of CART and C4. 5 can further improve the prediction accuracy than each single method. Furthermore,in consideration of the characteristics w hose features have high or low correlations w ith the response variable,mutual information w as introduced as an evaluation criterion for feature selection. M utual information makes great contribution to classification effect of random forest algorithm,and recognition rate of defects of hot-rolled strips is obviously improved by using three strategies.更多还原