GTM在冷轧带钢颤振监测与识别中的应用  

Monitoring chatter in cold mill based on generative topographic mapping

作　　者：廖广兰[1] 陈勇辉[1] 史铁林[1] 

机构地区：[1]华中科技大学机械科学与工程学院

出　　处：《华中科技大学学报（自然科学版）》2003年第2期64-66,共3页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：国家重大基础研究项目 (G1 9980 2 0 3 2 0 ) ;湖北省自然科学基金资助项目 ( 2 0 0 0J1 2 5 )

摘　　要：研究了GTM算法在冷轧带钢颤振监测识别中的应用 ,结果表明 ,通过对轧制过程的工艺参数或振动特征的聚类训练 ,它能把正常数据和颤振数据分别映射到二维平面的不同输出区域 ,较好地将两种轧制状态聚类分开 ;输入数据映像点在平面上的轨迹变化趋势正确而直观地反映了轧制状态的变化 ,便于及时监测识别出颤振的发生 。Generative Topographic Mapping (GTM) a new unsupervised clustering algorithm, was described and applied to monitoring chatter in cold rolling mill. The experimental analyses demonstrate that input data can be projected into different areas in a two dimensional space using GTM in accordance with their states by training rolling parameters or vibrations features measured from work rollers, chatter and normal condition distinguished clearly. With the trace of image points in the two dimensional space, the variation of rolling conditions can be observed visually and correctly, and occurrence of chatter can be monitored in time. The rolling productivity can be increased and rolled strip quality improved.

关 键 词：冷轧带钢 颤振监测 颤振识别 GTM模型 轧制过程 GTM算法 非线性隐含变量模型 

分 类 号：TG335.56[金属学及工艺—金属压力加工] TP277[自动化与计算机技术—检测技术与自动化装置]