Data-Driven Iterative Learning Control for Continuous-Time Systems

We develop a data-driven iterative learning control design framework for continuous-time systems that does not require explicit or implicit identification of a system model. Using Chebyshev polynomial orthogonal bases, we show that all system trajectories can be characterised from sufficiently rich input/output data. Using this crucial result we develop a data-driven version of the model-based norm-optimal iterative learning control algorithm, and provide a computationally efficient implementation thereof. We rigorously analyse the convergence properties of the resulting design and also present a numerical example to illustrate its effectiveness.