Manual ILC iterations

To perform ILC on a physical system, we need to repeatedly apply an ILC signal to our system under test, record the result, and compute a new ILC signal. The function ilc that is used in the examples in this documentation performs all these steps internally, but using a simulation model of the system. To perform these steps manually, we need to do the following:

Initialize the algorithm using the function init. Also choose an initial ILC signal, typically all zeros, but may also be chosen as the result of a few ILC iterations on a simulated system.
Perform an experiment on the system under test and record the resulting output $y$
Compute the tracking error $e = r - y$
Compute a new ILC input signal $a$ using the function compute_input
Repeat steps 2-4 until the desired performance is achieved.
Enjoy sweet tracking performance, maybe also tell your mother about it.

The function workspace = init(prob, alg) takes the specification of the ILC problem and the chosen algorithm and returns a workspace object with problem-specific things the algorithm needs to run.

The function a = compute_input(prob, alg, workspace, a, e) takes the problem, algorithm, the workspace object created by init, the previous ILC input signal $a_k$ and the tracking error $e = r - y$ and returns a new ILC input signal $a_{k+1}$.

Docstrings

IterativeLearningControl2.init — Function

workspace = init(prob, alg)

Initialize the ILC algorithm. This function is called internally by the funciton ilc but manual iterations require the user to initialize the workspace explicitly.

source

IterativeLearningControl2.compute_input — Function

compute_input(prob, alg::ILCAlgorithm, workspace, a, e)

Compute the next ILC input using the learning rule

Arguments:

alg: The ILC algorithm
workspace: A workspace created by calling init on the algorithm: workspace = init(prob, alg).
a: Previous ILC input
e: Error r - y

source