Interpretable Predictive Modeling and Multi-Objective Optimization for Laser Cladding Via Enhanced WOA-ESN and Hybrid Quantum MOGWO

To address the strong nonlinear coupling between laser cladding parameters and coating quality for Stellite 12 on 2205 duplex stainless steel, this study proposes an interpretable learning and multi-objective optimization framework for process prediction and inverse parameter design. An Echo State Network optimized by the Enhanced Whale Optimization Algorithm is constructed to predict cladding efficiency, dilution rate, and forming coefficient, and SHAP analysis is employed to quantify the marginal contribution of each parameter and enhance model interpretability. For inverse optimization, a Hybrid Quantum Multi-Objective Grey Wolf Optimizer is developed, and a density-weighted ideal point method is applied to select a robust compromise solution from the Pareto front. The proposed model achieves high prediction accuracy with R² greater than 0.98 and outperforms conventional approaches. Experimental results confirm that the optimized parameters increase cladding efficiency by 17.39