Intelligent predictor model for critical current density of MgB2 superconducting bulks

The critical current of superconducting materials, such as magnesium diboride (MgB2) bulk superconductors, is a key parameter influencing their performance in various applications, including magnetic field shielding, MRI, and Maglev systems. Spark plasma sintering (SPS) is one of the most efficient methods to fabricate high-quality MgB2, significantly saving fabrication time and controlling grain growth. The fabrication conditions, including temperature, pressure, and dwell time, can affect the critical current density. Traditional methods for estimating critical currents are time-consuming and costly. This study explores the use of advanced artificial intelligence (AI) techniques to develop accurate and efficient models for predicting the critical current in MgB2 bulks with respect to 10 different influential fabrication properties and physical conditions. By using AI algorithms such as Gaussian process regression, extremely gradient boosting, and generalized regression neural network (GRNN) an extremely high accuracy in predictions against the actual experimental data was achieved. By defining and studying the extrapolation scenario, this study goes beyond of simple AI-based estimator model that performs well only within the training range of data. The developed AI models not only reduce the need for extensive experimental campaigns but also offer real-time prediction capabilities, paving the way for faster advancements in the research and development of superconducting technology. Overall, GRNN model demonstrated a good performance for both interpolation and extrapolation tasks with an R-squared of 0.999958 and 0.99521, respectively.