【摘要】 基于板坯连铸结晶器温度在线监控系统,在实现结晶器温度及其变化速率"热成像"的基础上,借助阈值分割算法对温度变化的可疑区域进行提取,并采用八连通判别算法对异常区域进行区分和标记,开发基于计算机图形学的结晶器黏结漏钢可视化预报方法。以此为基础,从异常区域的位置、扩展、移动以及形状等方面,归纳和提炼结晶器黏结的共性特征,并与伪黏结进行区分。实验结果表明:基于图像处理的黏结漏钢预报方法,能够将伪黏结有效剔除,直观呈现异常发生和传播的全部过程及其典型特征,为异常在线诊断和准确预报提供先进手段,对于促进连铸生产的智能化、可视化控制水平具有积极意义。更多还原

【Abstract】 Based on the mould temperature on-line monitoring system for slab continuous casting, mould temperature and its velocity thermographs were realized. The suspicious temperature regions in the thermograph were extracted and divided by virtue of threshold segmentation. And the abnormal zones were determined with the eight connected components labelling algorithm. A visual detection method for mould sticker breakout was developed based on computer graphics. Then based on it, the common characteristics of sticker breakout were collected and compared with the false sticker breakout on the abnormal zone location, extension, horizontal move and shape, etc. The results show that the sticker breakout prediction method based on image processing can present the classical formation and propagation characteristics of sticker breakout and distinguish false sticker breakout intuitively. It also provides an advanced means to detect and predict the abnormity, and has a positive meaning to the intelligent and visual monitoring system for continuous casting production.更多还原