Design and Implementation of System for Generating MOFs for Hydrogen Storage in Hydrogen-Fueled vehicles

Abstract

New energy vehicles replace nonrenewable energy such as gasoline with renewable resources. On the one hand, it effectively reduces the use of nonrenewable energy and protects the natural environment. On the other hand, it improves the atmospheric environment. Because hydrogen energy has the characteristics of high efficiency and energy saving, the use of hydrogen energy has become one of the directions for the development of clean energy. Metal-organic framework (MOF) has been widely studied in the field of gas adsorption due to its porous structure and large specific surface area. In this paper, we propose a system for generating MOF based on Monte Carlo Tree Search (MCTS). By improving the activation function in the gated recurrent unit (GRU) structure, the convergence speed and accuracy of the neural network can be improved. The improved GRU is used as a policy network to guide MCTS to generate MOFs with superior hydrogen adsorption performance. The improved GRU has an accuracy of 90.31% on the SMILES string dataset, which is 1.19% higher than the accuracy of the traditional GRU; For specific metal nodes and topologies, the system can generate MOFs with larger hydrogen adsorption capacity than the experimentally synthesized MOF materials.