Temperature prediction of high-temperature and high-enthalpy plasma generators based on machine learning.

Inductively coupled plasma (ICP) is widely used in aerospace engineering and material processing for generating high-purity, high-temperature airflow, crucial for applications like thermal protection and plasma stealth. A scientific device developed at Xidian University utilizes ICP technology to simulate plasma sheath characteristics and enables high-resolution plasma diagnostics. This study applies machine learning (XGBoost algorithm) to predict temperature characteristics of quartz tubes in plasma generators, focusing on preventing overheating and improving system stability. Generated models are utilized to predict the temperature extremes and the heating rates of the quartz tube in various experimental conditions. By analyzing key features such as argon and air intake rates, intake durations, and coil voltage, this study demonstrates that machine learning delivers highly accurate predictions (R2 = 0.90 for temperature peaks and R2 = 0.82 for heating rates). This work also quantitatively emphasizes the significance of voltage and air intake parameters in determining the temperature characteristics of plasma generators and providing key insights for optimizing system performance and improving experimental standardization, ensuring their stable and long-lasting performance. [ABSTRACT FROM AUTHOR]