Multi-resolution Wavelet Method for Roll Eccentricity Compensation in Strip Rolling Mills

In order to compensate the roll eccentricity in hot strip rolling process, a multi-resolution wavelet transform method was proposed. This wavelet method has quiet good localization characteristic in both time domain and frequency domain. To eliminate frequency aliases inhered in Mallat algorithm, an anti-aliasing algorithm combining fast Fourier transform (FFT) and inverse fast Fourier transform (IFFT) was introduced. Its implementation was described in detail. The roll eccentricity component was extracted and reconstructed from the integrated rolling force signal based on this proposed wavelet method. A controller added with multi-resolution wavelet analyzer and roll eccentricity compensator was designed. Results from an experiment illustrate that this new method could greatly restrain the inverse effect of eccentricity and improve the thickness control accuracy.