Frequency-Domain Iterative Learning Control With Applications To Injection Molding Machines

In general, the tracking error is inclined to be "convergent-divergent-convergent", i.e. not monotonic decay required in practice, when in the time-domain-based iterative learning control (ILC). Then, a frequency-domain ILC is presented to prevent bad learning transient. However, it is difficult to require the sufficient conditions of convergence derived in frequency domain to hold for all frequencies. This thesis concentrates on widening the learnable frequency band so that much more frequency components entering the learning should be in this band and small tracking error allowed is ensured. Case studies are carried out to demonstrate the effectiveness of the proposed learning control schemes for the multi-level injecting velocity control of super injection molding machines. Several results achieved in this thesis are as follows:(1) Robust ILC with disturbance rejection is considered for uncertain continuous-time systems. The convergence condition is proven to be the same as the well-known robust performance condition. Furthermore, the method of how to select the cut-off frequency of performance weighting function is provided, and then the controller is designed based on existing robust control theory.(2) A frequency domain design is presented for the discrete anticipatory learning control. Convergence conditions are derived in terms of two design parameters, the lead time and the learning gain. For...