Roll eccentricity is an important factor causing thickness variations during hot strip rolling and might define the limit of strip thickness control accuracy. An improved multi-resolution wavelet transform algorithm was proposed to compensate for the roll eccentricity. The wavelet transform method had good localization characteristics in both the time and frequency domains for signal analysis; however, the wavelet method had a frequency-aliasing problem owing to the less than ideal cut-off frequency characteristics of wavelets. This made its component reconstruction of an inaccurate signal. To eliminate inherent frequency aliases in the wavelet transform, fast Fourier transform (FFT) and inverse fast Fourier transform (IFFT) were combined with the Mallat algorithm. This synthesis was described in detail. Then, the roll eccentricity component was extracted from rolling force signal. An automatic gauge control (AGC) system added with a multi-resolution wavelet analyzer was designed. Experimental results showed that the anti-aliasing method could greatly restrain the inverse effect of eccentricity and the thickness control accuracy was improved from ±40 μm to ±15 μm.
[1]
Liu Xiang-hua.
AGC System With Adaptive Threshold Method of Wavelet Transform
,
2007
.
[2]
Stéphane Mallat,et al.
A Theory for Multiresolution Signal Decomposition: The Wavelet Representation
,
1989,
IEEE Trans. Pattern Anal. Mach. Intell..
[3]
Min Huang,et al.
Roll eccentricity compensation control for strip rolling mills based on wavelet packet de-noising theory
,
2004,
Fifth World Congress on Intelligent Control and Automation (IEEE Cat. No.04EX788).
[4]
Jianguo Yang,et al.
AN ANTI-ALIASING ALGORITHM FOR DISCRETE WAVELET TRANSFORM
,
2003
.
[5]
Andreas Kugi,et al.
Neural network for identification of roll eccentricity in rolling mills
,
1996
.