Speaker
Description
Lutex is a high-performance irradiation facility providing the world’s largest testing area (400 × 700mm2). It delivers light-ion beams with very high fluxes at 300keV/u (∼5000km/s) for H+ up to 4×109 protons/cm2/s. This enables rapid reproduction of extreme exposure conditions and allows for short time qualification of a material lifetime. For comparison, a typical solar storm reaches only 3keV/u (∼600km/s) and 0.2 protons/cm2/s.
Operating a continuous accelerator such as Lutex requires achieving beam stability conditions on energy and current over a long period of time tailored to experimental needs. The difficulty resides in the change of a large number of input parameters, in their nonlinear interactions and in the variability of irradiation conditions.
Pantechnik is developing a Machine Learning model designed to facilitate beam tuning, to optimize beam quality, and to predict parameters required for production of continuous beam. The model exploits data collected from plasma generation, extraction, transport, separation, and dose measurements to learn the relationship between machine settings and resulting beam performances. After data preprocessing and feature extraction, several approaches, including Bayesian optimization and nonlinear regression methods, will be benchmarked to predict and optimize the beam current as a function of operating parameters.
| Paper status | Proceeding files received |
|---|