Segmentation filtering iterative learning control method of motor servo system

Provided is a segmentation filtering iterative learning control method of a motor servo system. The method comprises steps of firstly determining learning law convergence frequency domain scope is 0-f when a zero-phase filter, i.e., Q(z)=1, is not introduced; then performing reference trajectory track of the primary iteration-free loop so as to obtain an error signal e0(t), applying the EMD algorithm to decompose the e0(t) into limited number of intrinsic mode functions, and obtaining time sequence characteristics of the error signal e0(t) via the Hilbert conversion; then using frequency f as the segmentation reference frequency; segmenting the error signal e0(t), and distinguishing time segments where instant frequency of the error signal e0(t) is higher than the f and lower than the f ; designing cut-off frequency of a zero phase according to the Hilbert time sequence characteristics of the e0(t); in the time segment where the instant frequency of the error signal is higher than the f , f =f +1/2(f -f ) is adopted as the cut-off frequency of the filter, wherein fmax represents the highest frequency in this segment; in other time segments, f is adopted as the cut-off frequency of the filter; and at last, applying the designed segmentation zero phase filter to the following iterative learning control processes.