基于改进小波去噪和EEMD的轧辊偏心提取与补偿  

Roll eccentricity extraction and compensation based on improved wavelet denoising and EEMD

作　　者：魏立新[1] 冀晓凯 范锐 孙浩[1] WEI Li-xin, JI Xiao-kai, FAN Rui, SUN Hao(Key Lab of Industrial Computer Control Engineering of Hebei Province, Yanshan University, Qinhuangdao 066004, China)

机构地区：[1]燕山大学工业计算机控制工程河北省重点实验室

出　　处：《塑性工程学报》2018年第5期298-306,共9页Journal of Plasticity Engineering

基　　金：河北省自然科学基金资助项目(F2016203249)

摘　　要：为准确提取轧辊偏心信号,进而实现偏心补偿控制,提高冷轧机的厚度控制,提出了将一种改进小波阈值去噪和EEMD相结合的偏心信号提取方法。该方法结合了小波的强去噪性以及EEMD的抗模态混叠的优点,采用一种含参数的可变阈值函数,在阈值选择时通过人工蜂群优化算法自适应确定最优阈值。利用改进的小波阈值法对轧辊偏心扰动信号进行去噪预处理,然后经过EEMD将信号分解,提取表征偏心信号的特征模态函数,将重构的偏心信号补偿到冷轧机系统中。最后,通过仿真实验表明,此方法能有效补偿轧辊偏心,且所得补偿效果明显优于小波算法的补偿效果。In order to accurately extract roller eccentricity signal,realize eccentricity compensation control,and improve thickness precision of cold rolling mill,an eccentric signal extraction method was proposed by combining the improved wavelet threshold denoising and the ensemble empirical mode decomposition（ EEMD） approaches. This method combined the strong denoising of wavelet and the anti-modal aliasing of EEMD. A variable threshold function was constructed to determine the optimal threshold by an artificial colony optimization algorithm. The improved wavelet threshold method was used to denoise the rolling force signal,and the characteristic modal function for representing the eccentricity signal could thus be extracted by decomposing the signal with the EEMD. Then,the reconstructed eccentricity signal was compensated into the rolling mill system. Via simulation experiments,the proposed method presents a better compensation capability than the wavelet algorithm,and can effectively compensate the roll eccentricity.

关 键 词：小波去噪 人工蜂群算法 集合经验模态分解 偏心信号提取 偏心补偿 

分 类 号：TG331[金属学及工艺—金属压力加工]