Study of Liquid Lithium Coolant Interaction Based on BP Neural Network Optimized by Genetic Algorithm

Liquid lithium provides a viable alternative to traditional solid divertors and is one of the important choices for plasma facing materials in magnetic fusion devices. The liquid lithium coolant interaction in the accident conditions is a great threat to the safety of the devices. The prediction of explosion strength of liquid lithium coolant interaction is the key in the assessment of related accidents in fusion reactors. It is a kind of complicated nonlinear relation between explosion strength and its influencing factors, including the mass of lithium, initial lithium temperature and initial coolant temperature. Therefore, an optimized BP neural network model for predicting the explosion strength has been developed and it has been tested by the experimental data. The genetic algorithm is applicable to optimize the weights and thresholds of the BP neural network to obtain better prediction results. The comparison between the prediction results of the optimized BP neural network and the original one shows that the optimized prediction model is more accurate and efficient. It provides an optimized method for the evaluation of explosion strength of the liquid lithium coolant interaction.