Predicting the Initial Discharge Capacity of AB5-Based Hydrogen Storage Alloy Using Artificial Neural Network

Back-propagation artificial neural network was developed to predict the initial discharge capacity of AB5-based hydrogen storage alloy. Leave-one out method was used to train the ANN model. Test results showed that the prediction performance of the ANN model is satisfactory: the scatter dots distribute along the 0__45deg diagonal line in the scatter diagram, the values of statistical criteria are 11.4407 mAh/g(MSE), 4.78% (MSRE), and 1.6413 (VOF) respectively. Moreover, the ANN model was used to analyse the quantitative effects of chemical elements on the initial discharge capacity, results showed that inverse-parabola relationship exists among C0 and La content, and parabola relationship exists among C0 and Ce and Nd content, and the C0 value decrease with the increase of Pr content.