Prediction and Application of Critical Current in 2G HTS Conductors Based on GA-BP Algorithm

Abstract

Critical current, as a key parameter distinguishing superconductors from ordinary conductors, determines the stability of superconducting devices and systems during operation. To accurately assess the critical performance of superconductors, a genetic algorithm (GA)–optimized back-propagation (BP) neural network model is introduced and used in this paper to predict the critical currents of second-generation high-temperature superconductors (2G HTS). Firstly, a staged optimization is carried out for the neural network structure and hyper-parameters, and the GA-BP model is established by constructing different fitness functions. Next, the prediction accuracy and generalization ability of the GA-BP model are validated by comparing the relative errors between the predicted values from two models in the 180° to 240° anti-angle region. Eventually, the critical current at a specific temperature is predicted by the GA-BP model, and the corresponding J_c0 is calculated for the finite element calculation of superconducting strips. The calculation results show that the relative error between the maximum current density obtained based on the predicted J_c0 and the experimental J_c0 is only 0.907%, indicating that the model can be used to accurately determine the operating state of the superconducting equipment.