Toward the application of inertial fusion energy, a comprehensive comparison of different fusion materials was made. Using the upgraded multi-fuel fusion package of the radiation-hydrodynamic code MULTI-IFE, datasets of fusion reactions for different fusion fuels were established.It was demonstrated that the D–3He reaction has the potential to achieve a fuel energy gain greater than 100. Taking advantage of transfer learning, the pre-built deep neural network of D–T fuel was successfully translated to other materials, including D–3He and D–D fuels. Considering the generation of tritium and helium via D–D reactions, both the D–T and D–3He fuels would be acceptable for the upcoming clean and economic fusion power plants.

transfer learning,inertial confinement fusion