Use of Machine Learning for the Measurement of Wire Sag in a Pulsed Wire Set-up

The pulsed wire method (PWM) belongs to the family of well-established measurement techniques for undulator magnetic field characterization. Wire sag is a deformation caused by the weight of the wire under tension, which introduces measurement errors. Wire sag measurements of the enameled copper wire, with a thickness of$500~\mu $m (25 swg), are done for a pulsed wire setup. The wire sag is measured for various wire lengths in the pulsed wire setup with different tensions in the wire. The measured data for the wire sag is compared with the theoretical predictions of wire sag. Machine learning (ML) models, including a random forest regressor and a multilayer perceptron (MLP) regressor, were also employed to predict sag for unmeasured tensions. The MLP model performed better, reducing deviations between experimental and theoretical sag values. This method improves sag correction, making PWM setups more reliable for undulator field measurement to find the sag for unmeasured tensions to improve the correction with theoretical sag, which can be used for correction in undulator field measurement.