Removing non-repetitive disturbances in iterative learning control by wavelet filtering

The tracking performance of systems that perform repetitive tasks can be significantly improved using iterative learning control (ILC). During successive iterations, ILC learns a high performance feedforward signal from the measured tracking error. In practice, the tracking error consists of both a repetitive part which is equal every iteration and a non-repetitive part which varies every iteration. ILC can only compensate for the repetitive part, the non-repetitive part limits the achievable performance of ILC. In this paper, a wavelet based filtering method is presented which identifies and removes the non-repetitive part of the tracking error by a comparison of two error realizations for each iteration of ILC. The filtered error signal is used as input for the learning scheme of ILC. Simulations and experiments show that the wavelet filtering method improves the performance of ILC, resulting in a smaller tracking error and in a learned feedforward signal that contains significantly less non-repetitive disturbances