Frequency Domain Based Conditions for Determining Convergence Learning Parameters in Linear Iterative Learning Control

Tóm tắt

This article proposes four sufficient conditions to determine the convergence learning parameters in iterative learning control of linear batch processes. These conditions are established in frequency domain with transfer functions of elementary linear learning functions of P-, D-, PD- and PID-Type, instead of their state space models as usual. Hence, they can overcome all conservative difficulties occurred by using conditions created in time domain. To obtain these conditions in frequency domain, first an overall sufficient condition belonging to input-output mapping is created, and then realized it particularly in frequency domain for four different linear learning functions by using their transfer function. The obtained conditions in frequency domain are expressed 
in algebraic inequality of matrix norm, so they are very convenient in use. To illustrate the applicable ability of proposed conditions in various practical applications some numerical simulations had been carried out in the paper. Obtained simulation results authenticate the advantage of these conditions.