Abstract:

The paper introduced the working principle of flatness signal measurement on pneumatic flatness measuring roller, developed a flatness pattern recognition system by virtual instrument LabVIEW, which was made up of polynomial least square method, orthogonal polynomial least square method, fuzzy pattern recognition based on Euclidean distance. By measuring on-line tension signal of cold mill aluminum stripe on pneumatic flatness measuring roller, three flatness pattern recognition methods were applied respectively. Test shows the flatness pattern recognition system possesses good stability and validity, and good real time measurement and graphics interface communion capability of system provide significant foundation for flatness closed loop control.

Key words: pattern recognition, fuzzy, orthogonal polynomial, least square method, virtual instrument

摘要：

介绍了气动板形检测辊的板形信号检测原理，详细阐述了普通多项式最小二乘法、正交多项式最小二乘法和基于欧氏距离的模糊模式识别法，利用虚拟仪器LabVIEW软件开发了基于上述三种方法的板形模式识别系统。通过气动板形仪在线测试了300型可逆冷轧机板形张力信号，并对张力信号进行了模式识别。实验表明：该板形模式识别系统具有良好的稳定性和有效性，良好的图形交互界面和实时测试性能为板形闭环控制提供了方便。

关键词: 模式识别；模糊；正交多项式；最小二乘法；虚拟仪器