Analysis of anode temperature stability in hollow cathode vacuum arc brazing under multi-parameter coupling based on SR-MLP

During hollow cathode vacuum arc brazing, the anode temperature is prone to instability due to the interplay of multiple parameters, leading to fluctuations in joint temperature and potential localized melting of the base material. To address this issue, this study investigates anode temperature stability under multi-parameter coupling. First, a multiphysics mathematical model integrating fluid flow, plasma, and temperature fields was established, and numerical simulations of the temperature field were performed using COMSOL Multiphysics. Results reveal that anode temperature positively correlates with argon flow rate and current size, while negatively correlating with bipolar spacing and cathode tube radius. Subsequently, to quantify anode temperature stability, a definition of "anode temperature stability" was proposed, and a theoretical sample data set was established under various process parameter perturbations. A combined Multilayer Perceptron (MLP) and Symbolic Regression (SR) approach was employed to construct a model describing anode temperature stability. Descriptive model parameters were determined through regression analysis. Finally, by substituting 10 sets of actual welding parameters from the hollow cathode vacuum arc brazing site into the descriptive model, the validation results indicate: The model's Mean Absolute Error (MAE) was 0.83, Root Mean Square Error (RMSE) was 0.95, and Mean Absolute Percentage Error (MAPE) was 3.49%, indicating high prediction accuracy. The proposed stability model offers a robust theoretical foundation and practical support for closed-loop intelligent control of the hollow cathode vacuum arc brazing process.