10th International Conference on Surface Engineering (ICSE 2026)

Hydraulic machinery flow passage components suffer erosion damage under high-speed sediment-laden flow. Most studies treat composition and process parameters separately, overlooking their coupled effects on macroscopic properties and microstructure. This study proposes a synergistic optimization framework integrating machine learning and multi-objective optimization for laser cladded Al_xCoCrFeNi_y high-entropy alloy coatings. Inputs include laser power, scanning speed, powder feed rate, Al and Ni contents. Outputs are microhardness, height, dilution rate, aspect ratio, and Cr-segregated region. BP achieves the highest accuracy for microhardness prediction (R²=0.951, RMSE=14.3), while GA-BP exceeds R²>0.9 for the other four targets. Garson analysis reveals that Al and Ni govern microhardness and Cr-segregation region, while process parameters control geometry, confirming the necessity of incorporating both composition and process variables in the framework. NSGA-II with TOPSIS is used for inverse multi-objective optimization, yielding four coatings with distinct microstructures. Notably, the microhardness-prioritized Specimen 2# (Al=14.5 at.%, Ni=19.1 at.%), with the highest Al content, promotes the formation of a single-phase BCC structure. Its moderate Ni content drives continuously distributed network-like Cr segregation, which serves as the matrix for BCC phase, providing uniform microhardness contribution. Specimen 2# exhibits the highest microhardness and the best erosion resistance, with culmulative mass losses of only 11.5%, 9.6%, 9.5%, and 7.0% of the substrate at 30°, 45°, 60°, and 90° impact angles, respectively. Microhardness governs the erosion mechanism, suppressing micro-cutting at low angles and enhancing fatigue resistance under normal impact. This study provides a guidance for composition-process synergistic optimization of high-performance erosion-resistant coatings via machine learning and multi-objective optimization.

Machine learning, Multi objective optimization, Laser cladding, AlCoCrFeNi, Erosion resistance