Simulation of hard X-ray time evolution in the stable region of plasma tokamak by using the NARX-GA hybrid neural network

The time evolution of hard X-ray has been simulated using the NARX-GA hybrid neural network in the stable region of the plasma tokamak. Loop voltage and hard X-ray measured by the tokamak diagnostics tools were selected as network inputs. The NARX network has been trained using the Genetic Algorithm (GA) and the time evolution of the hard X-ray up to ۵۰۰ μs (MSE = ۴.۱۳ × ۱۰-۵) is accurately simulated. Increasing the confinement time is the particular purpose of applying tokamak to produce energy through fusion. The real-time application of this methodology brings us closer to this goal. Hard X-ray prediction can prevent plasma energy reduction. It can also reduce the severe damage caused by runaway electrons (RE) colliding with the tokamak wall. Early prediction of hard X-ray time evolution is critical in attempting to mitigate the REs potentially dangerous effects.

Keywords:

Methodology , Hard X-ray , Runaway electrons , NARX-GA network