Command Shaping for MIMO Nonlinear Systems Using Iterative Learning Control With Application to an RTP System
Abstract Input command shaping for temperature control of fast-ramp RTP systems is investigated from an open-loop-input point of view, i.e., for a given desired temperature recipe a set of lamp command profiles is determined such that the resulting set of measured temperatures approaches the desired recipe as closely as possible. Because of the inherent nonlinear behavior of RTP systems, a command shaping method has been developed that iteratively modifies the optimal linear commands to compensate for the nonlinearities. This method, which has been derived from Iterative Learning Control (ILC), shapes the input commands iteratively so as to minimize the two-norm between a desired output trajectory and the simulated current output trajectory. The technique is applicable to MIMO systems and can handle constraints on the input commands. Application of this method to a fast-ramp oxidation (RTO) and fast-ramp spike anneal (RTA) process for a model of a generic RTP system demonstrates its usefulness for nonlinear systems.