Study For The Key Technology Of The Roll Eccentricity Compensation Control In Cold Strip Rolling Mill

The research of roller eccentricity is the important part of the research of the control technology of high-quality strip rolling. With the increasing request of the every walk of life to strip quality, influence affected by the eccentricity in strip quality has no longer been neglected and becomes the hotspot problem in the study of realm of modern strip rolling. Although it has had a lot of achievements in roller eccentricity research domain in our country, there is a relative gap compared with the developed countries. And there are many questions both in practical appliance and theory.This paper is based on the full investigation and practices; the research aiming at a few key techniques which are baffling the improvement of eccentricity compensation was carried on deeply. The corresponding conclusion and a new method forward were presented, and some simulation experiments were carried on.Firstly, based on the completely familiarity with the cold rolling control system and the principle of the AGC, we propose dynamic structure of the AGC system with the eccentricity compensation and simplified structure of eccentricity compensation control. For the existed problem that applying the signal of eccentricity compensation to the control system directly, we propose the efficient method of determining of the eccentricity compensation signal, which is based on the relation between the eccentricity compensation signal and the frequency feature of position closed loop system.Secondly, the limitation of the FFT (Fast Fourier Transform) in practice is discussed based on the basic theory of FFT. Combine the characteristic of the Hanning window in time domain and frequency domain, a new method that is used in the process of eccentricity signal extraction where the spectrum leakage occurred is offered, this method is named interpolation method of FFT based on the Hanning window.Lastly, we analyze the partial sway phenomenon of the eccentricity signal. This phenomenon is caused by the bias between the top backup roll and the bottom backup roll, and is very common in practice. We find a new least squares identify method based on the Givens transform, and we apply this method to identify the eccentricity signal, propose the algorithm of ordinary least squares and the algorithm of recursive least squares, both them can help the control system identify the eccentricity signal fast and efficiently.