Neuro-fuzzy and fuzzy modeling of laser alloying of steel $\mathrm{30Х\Gamma СН2А}$

The study focuses on developing neuro-fuzzy and fuzzy models for laser alloying of structural steel $\mathrm{30Х\Gamma СН2А}$ with chromium, based on numerical experiment results. Finite element modeling of laser alloying for steel $\mathrm{30Х\Gamma СН2А}$ was performed using APDL programming language, accounting for temperature-dependent thermophysical material properties and power densities, and employing a face-centered central composite experimental design. The experimental factors included time intervals corresponding to the durations of three laser pulse fronts and their peak power densities, while the response variables were maximum temperatures in the processing zone. The model quality was evaluated using statistical metrics ($RMSE$, $MAE$, $MAPE$, $R^2$) on a test dataset. The results demonstrate the superior predictive accuracy of neuro-fuzzy models over conventional fuzzy models for laser alloying parameters.