Phenomenological dynamics of coronal loops using a neural network approach Original Research Article

The objective of this study is to simulate the X-ray 2D-coronal dynamics using an artificial neural network — multilayer backpropagation algorithm with inputs of Energy Fragmentation Patterns obtained from Yohkoh images in soft and hard X-rays. Details of a single loop structure have been investigated for initial analysis. The images are spatio-temporal series showing the loop-top in Soft X-ray (SXR) and foot points in Hard X-ray (HXR). Using a square electron density gradient model, we have characterized the spatio-temporal loop dynamics concerning its twister-relaxation regime. The performance of this trained network model has been tested with classical image statistics applied to the Yohkoh data. In this paper we show preliminary results indicating that this technique can be useful for coronal dynamics analysis.