Abstract We present the first step towards an intelligent thin-disk multi-pass amplifier (I-TDMPA), by simulatively proving the ability of a neural network (NN) for the detection of misalignment. A digital twin of one of our experimental systems was implemented using a specifically developed ray tracing simulation. The digital twin was used to generate more than 200,000 sets of training data to train a neural network (NN) with the goal to be able to determine misalignment of the many optical elements of a multi-pass amplifier in a simulation model, solely by observing the beam. The predictions of the NN agree very well with the simulated misalignments, which confirms the suitability of artificial-intelligence methods to ensure a safe and stable operation of complex optical systems.
Bienert et al. (Wed,) studied this question.