基于Adaboost加权支持向量机的热轧板带弯曲性能质量预警  

Research on quality warning of strip bending performance based on Adaboost-weighted support vector machine

作　　者：周鹏 何飞 梁冰 徐科 ZHOU Peng;HE Fei;LIANG Bing;XU Ke(Engineering Technology Research Institute, University of Science and Technology Beijing, Beijing 100083, China;Collaborative Innovation Center of Steel Technology,University of Science and Technology Beijing, Beijing 100083, China)

机构地区：[1]北京科技大学工程技术研究院,北京100083 [2]北京科技大学钢铁共性技术协同创新中心,北京100083

出　　处：《中南大学学报（自然科学版）》2017年第10期2623-2628,共6页Journal of Central South University:Science and Technology

基　　金：国家自然科学基金资助项目(51204018);"十二五"国家科技支撑计划项目(2015BAF30B01);中央高校基本科研业务费专项资金资助项目(TW201711;FRF-TP-16-018A1)~~

摘　　要：针对热轧带钢弯曲性能质量监控与预警过程中因正常样本与异常样本的比例严重失衡而导致质量监控过程中预警不灵敏、异常检出率较低的问题,从数据层面和算法层面研究不平衡样本数据的质量预警问题,提出基于Adaboost加权支持向量机的热轧带钢弯曲性能质量预警方法。研究结果表明:采用该方法所得平均异常检出率提高至88.58%,误判率为0.63%。该方法具有较强的异常检出能力,能够为热轧板带生产过程的质量预警提供保障。Due to the data imbalance between normal and abnormal samples,the quality warning is not sensitive and the detection rate is low in the quality monitoring of strip bending performance.To solve this problem,different imbalance solutions to the abnormal detection from data level and algorithm level were studied.The quality warning method based on the Adaboost-weighted support vector machine method was proposed.The results show that the average value of fault detection rate is improved to88.58%,and the average value of false alarm rate is reduced to0.63%.The proposed method produces satisfying results in fault detection,which provides support for quality warning in hot rolling strip process.

关 键 词：带钢弯曲性能 质量预警 Adaboost加权支持向量机 

分 类 号：TG335.5[金属学及工艺—金属压力加工]