Designing iterative learning controllers via noncausal filtering

Abstract

It is advocated in this chapter that a PID-autotuning alike ILC design procedure is important for ILC to be acceptable by industry. This chapter has made some efforts in this direction by using the idea of noncausal FIR filtering techniques. The proposed learning scheme is very simple, straightforward and yet not model based. Convergence analysis has been given in detail for a class of zero phase filters. ILC design steps are given explicitly which is close to actual application situations. It is shown that the design task of ILC can be reduced to tuning two parameters: length of the filter and the learning gain. Some practical considerations in the parameter tuning are also outlined. Also, a limit on the ILC convergence rate has been established.