摘要
针对轧钢生产过程中在线故障检测和故障诊断的问题,提出一种基于多核学习算法的钢铁生产轧钢过程在线故障检测模型.首先针对学习样本建立核主成分分析与支持向量数据域描述模型,然后基于T2、Q统计量,以及数据域描述包络情况对轧钢过程进行初步识别,最后构建基于多分类多核最小二乘支持向量机预测模型,对初识结果进行细分类,识别故障级别.利用上述模型对轧钢加热炉故障和机组故障进行了试验.结果表明,该方法能有效检测钢铁生产轧钢过程的故障.
In order to solve the difficult problems of failure detection and diagnosis in steel rolling process,we propose an online failure detection model in rolling steel production process based on multi -kernel learning the-ory.First,for learning samples,a kernel principal component analysis and support vector data description model is built.And then,whether there is failure in the rolling process is detected simply by T2 ,Q statistics and data domain description enveloping surface.Finally,the least squares support vector machine failure prediction model is constructed based on multi-classification multi-kernel,which classifies and identifies the failure accurately. The above model is utilized in experiments for steel rolling heating furnace failures and assembling unit failures, the results showing that the method can effectively detect failures in the process of rolling steel production.
出处
《昆明理工大学学报（自然科学版）》
CAS
2015年第4期63-71,99,共10页
Journal of Kunming University of Science and Technology(Natural Science)
基金
国家自然科学基金项目(61175068
61163004)
关键词
轧钢过程
在线诊断
故障检测模型
多核学习
steel rolling process
on-line diagnosis
failure detection model
multi -kernel learning
