The analysis of inertial confinement fusion (ICF) plasma diagnostic pictures is crucial for fusion energy research. In this paper, a method based on the combination of deep reinforcement learning and computer vision techniques is proposed for the analysis of ICF plasma diagnostic pictures. The method first preprocesses the images using computer vision techniques and then uses deep reinforcement learning to classify and recognise them. The physical quantities are closer to the theoretical values using the new method, which is more instructive for the experiments. For example, at the radiation temperature, the obtained values are increased by 20-70 eV, and at the electron and plasma temperatures are close to the theoretical 5 KeV. at the same time the neutron yield is increased by a factor of 10. The experimental results show that the method has high accuracy and efficiency in the analysis of ICF plasma diagnostic pictures, and can effectively assist fusion research.
|