Study On The Method Of Angle Domain Analysis And Active Compensation For Roll Eccentricity

With the rapidly development of society, industry have an increasing demanding on thickness precision of the plate which is the main iron and steel products and take an important part in the national economy. The roll eccentricity which is caused by irregular shape of the roll body and poor assembly accuracy is a more complex periodic motion. A great deal of experiments shows that the higher the accuracy of rolling, the precision effect of roll eccentricity will be greater.Cold rolled strip control system is a multi-variable control process. Roll eccentricity are reflected in the various variables, which is difficult to extract the eccentricity signal accurately from a single information source. Our research adopted the roll eccentricity analysis method based on thickness difference tracing, extracted accurate eccentric according to comprehensive data containing the roll eccentricity signal sequence and conducted compensation research.Firstly, we analyzed the reason of roller eccentricity produces and the influence of rolling mill thickness and established a comprehensive data contains the roll eccentricity sequence, which provide conditions for the extraction of eccentric signal. As the rolling speed change frequently in the rolling process, a sequences transition method was put forward, which can transfer the time series into strip length sequence, meanwhile, eliminate the influence of speed fluctuation.Secondly, the algorithm of the time frequency analysis of thickness difference tracing was introduced. Because the resolution of the wavelet analysis is poor when deal with the high frequency detail, we proposed wavelet packet for the roll eccentricity signal denoising. According to wavelet packet frequency interleaving phenomenon, we summarized the reasons and obtained the right corresponding band relationship.Thirdly, based on experiments conducted in a school of 300 rolling mill and analysis of a large amount of historical data on the mill produced, we utilized wavelet packet analysis to obtain the eccentricity signal,compared with the eccentricity signal extraction of no-load pressed and verified the accuracy of the eccentric signal.Finally, according to the phase position, we took advantage phase information contained by the rolling force signal to correct the eccentricity signal source and verified by simulation that this method can accurately locate the eccentricity signal. Aiming at the problem that traditional PID control static error can eliminate the periodic signal, we proposed a new compound control method based on PID and repetitive control, the simulation results of which show that the composite control method is effective.