Study On The Extraction And Compensation Of Roller Eccentricity Signal For Cold Rolling Mill

With the improvement of the accuracy of strip rolling,it is particularly important to study the roll eccentricity that affects the quality of products.The roll eccentricity is reflected in the rolling force.In order to solve the roll eccentricity problem,the first task is to extract the eccentricity information from the rolling force signal accurately,and then compensate the extracted eccentricity signal to the control system,so as to reduce the fluctuation of the thickness of the exit plate.Aiming at the problem that roll eccentricity is periodic signal and is not easy to extract effectively,the research of eccentric signal extraction and compensation control is carried out.The main contents are as follows:Firstly,the theoretical basis of roll eccentricity is introduced.The reasons of roller eccentricity definition and eccentricity are analyzed,and the influence of roll eccentricity on rolling force and outlet thickness is further elaborated,the relationship between eccentric signal and rolling force signal is obtained,then a unified model of roller eccentricity signal is established according to the characteristic of eccentric signal.Secondly,for the eccentricity signal with complex interference,the eccentricity signal extracted by the empirical mode decomposition method in the rolling noise background is not accurate enough,and the ordinary wavelet algorithm is not good to eliminate the noise,and the frequency aliasing phenomenon is easy to appear.Therefore,a variable wavelet threshold function with parameter is proposed,and the optimal threshold is selected by artificial bee colony algorithm to realize the de-noising preprocessing of the roller eccentricity disturbance signal.Then the signal is decomposed by the Ensemble Empirical Mode Decomposition(EEMD),and the characteristics of the eccentricity signal are extracted.Modal function overcomes the frequency aliasing problem of wavelet algorithm.Simulation results show that this method can extract eccentric signals effectively.Then,the wavelet method is limited by the number of decomposition layers and the selection of the wavelet basis function.In the noise background,some effective information may be removed for the extracted eccentricity signal,resulting in the deviation of the extracted signal.The rotation invariant subspace technology has strong noise suppression and high frequency resolution.The frequency of the roller eccentricity signal is identified by the Total Least Square-Estimation of Signal Parameters via Rotation Invariance Techniques,TLS-ESPRIT),and then mining for the roll eccentricity signal.The improved bee colony algorithm is used to estimate the amplitude and phase of the signal.Finally,by comparing the improved method of wavelet denoising and EEMD decomposition and TLS-ESPRIT combined with improved artificial bee colony algorithm,the effectiveness of TLS-ESPRIT combined with improved artificial bee colony algorithm for extracting eccentricity signals is analyzed.Finally,based on the hydraulic press down system,the eccentricity signal is extracted from the collected rolling force signal.In view of the problem that the eccentric periodic signal is difficult to track accurately,the accurate eccentricity compensation control is carried out with the combination of PI and repetitive controller to achieve good compensation effect,and the method of roller eccentricity extraction and compensation is effectiveness.