System optimization of heat transfer performance of hydrogen storage bed based on backpropagation neural network-genetic algorithm

Abstract

ABSTRACT Seven structural parameters that affected the heat transfer performance were proposed as optimization variables to enhance the heat transfer and thermal reaction capacity of the thin double-layered annular zirconium–cobalt (ZrCo) bed of hydrogen metal reactor and to improve the hydrogen storage performance. We used the Taguchi method to conduct numerical experimental sampling and construct a three-dimensional model of the hydrogen absorption of a hydrogen storage bed. COMSOL software was used to simulate 36 models of the hydrogen absorption process and changes in the temperature and time of the hydrogen storage bed under different conditions were identified. A new hybrid method combining a neural network and the genetic algorithm was proposed by taking the hot-spot temperature of the bed and the cooling time when it was cooled to 300 K as the optimization objectives. The algorithm was implemented, and the relationship between the process parameters and the objective function was established. A model response analysis was conducted to improve the understanding of the behavior of the backpropagation model and to analyze the sensitivity of the parameters. This hybrid method was used to optimize the parameters to obtain an excellent hydrogen storage performance. The results showed that the predicted value of the neural network model was highly consistent with the numerical simulation results. The number of cooling tubes had the greatest impact on the heat transfer performance, and the optimal combination of the input parameters was obtained. When the optimized parameters were used to reach the target temperature, the cooling time was reduced by 78s, which provided guidance for the design and operation of hydrogen storage beds.