Predicting hydrogen storage capacity of metal–organic frameworks using group method of data handling

Abstract

Due to their unique properties, metal–organic frameworks have exhibited excellent performance for hydrogen storage purposes in the last decade. In this regard, model development to predict the hydrogen storage in metal–organic frameworks is of a vital importance for designing and developing of efficient processes based on these new synthetic material. The objective of the present study is to develop a new model to predict the hydrogen storage capacity in metal–organic frameworks. The group method of data handling-type polynomial neural networks is implemented as a soft computing approach for model building. As an advantage, only 40% of data points are used for model development and the rest of data (60%) are designated for testing of the model. The results show that the proposed model has reasonable accuracy in which the root mean square error for the proposed model is 0.28. The model can acceptably predict effects of surface area and pressure on hydrogen storage capacity of MOFs demonstrating good ability of the proposed model for tracing physically expected trend for hydrogen storage. Additionally, the leverage measure demonstrates that the proposed model is statistically acceptable and valid. It should be noted that an artificial neural network is also developed for comparison with GMDH-PNN model, in which the results confirm that both of the models have approximately same performance and accuracy. However, due to simple mathematical structure of GMDH-PNN, it is significantly more appropriate for engineering applications.