基于小波去噪的动态矩阵控制研究

为了消去动态矩阵控制（DMC）中系统输出所含的测量噪声，本文提出对一定长度的输出数据序列进行多层小波（尺度）分解，根据Stein无偏风险估计（SURE）法在每一层确定一个阈值，并用软阈值法对高频小波系数进行处理，然后将处理后的各层小波系数与尺度变换系数相加，重构出去噪后的当前时刻系统输出值．理论分析和实验都表明该方法在DMC中有效减小了静差，改善了控制精度．

Dynamic Matrix Control Based on Wavelet Denoising

In order to eliminate measurement noises of system outputs in dynamic matrix control (DMC) , the output measurements are decomposed by wavelet/scale multiresolution decomposition online. A threshold is chosen by means of Stein unbiased risk estimation (SURE) method in each scale, and high frequent wavelet coefficients are processed by soft thresholding. Then, current outputs of the system are reconstructed by adding the processed wavelet coefficents and the dimension transform coefficents. Theoretical analysis and simulation show that the method is effective to reduce steady state error and improve control accuracy in DMC.