Scholarlink

<p>The morphology of multi-track laser cladding coatings is influenced by various process parameters, including laser power, scanning speed, and powder feeding rate. Accurate morphology prediction is crucial for optimizing process parameters and enhancing coating performance. A novel prediction framework that integrates the Whale Optimization Algorithm (WOA) with the Random Forest (RF) algorithm (referred to as WOA-RF) is proposed to model the nonlinear relationship between the process parameters involved in laser cladding and the coating morphology. A central composite experiment design (CCD) with 4 factors and 5 levels was designed to carry out the laser cladding experiment, with the 17-4PH as the substrate and the 15-5PH as the cladding powder. The effects of the process parameters of laser cladding (laser power, powder feeding rate, scanning speed, and overlap rate) on the coating morphology characteristics (dilution rate, heat-affected zone area, and surface flatness) were analyzed. The relative importance of each process parameter was assessed by the random forest model. Based on the assessment results, a WOA-RF prediction model was established, and a multiple regression model and a random forest (RF) model were established for comparative analysis. The analysis reveals that the powder feeding rate exerts the most significant impact on the dilution rate, laser power is the key determinant of the area of the heat-affected zone, and the overlap rate predominantly affects surface flatness of the coating. The evaluation of parameter importance indicates that, under the conditions of this experiment, the overlap rate in the predictive model has a minimal contribution to the area of the heat-affected zone, and the laser power has a minimal contribution to the surface flatness of the coating. The WOA-RF model demonstrates superior performance in predicting morphological features across all evaluated metrics, outperforming both the RF model and the multiple regression model, with higher prediction accuracy. The proposed WOA-RF prediction method effectively captures the nonlinear relationship between process parameters and the morphology of multi-track coatings, which is helpful to determine the optimal process parameters to improve the coating morphology.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Method for Predicting the Morphology of Multi-track Laser Cladding Coatings Based on WOA-RF

  • 期刊论文

  • 发表日期：

    2025年3月13日

查看全文
  • Yanbin Du,
  • Xin Lei,
  • Hongxi Chen,
  • Qiang Liang,
  • Wensheng Ma,
  • Jian Tu

摘要

The morphology of multi-track laser cladding coatings is influenced by various process parameters, including laser power, scanning speed, and powder feeding rate. Accurate morphology prediction is crucial for optimizing process parameters and enhancing coating performance. A novel prediction framework that integrates the Whale Optimization Algorithm (WOA) with the Random Forest (RF) algorithm (referred to as WOA-RF) is proposed to model the nonlinear relationship between the process parameters involved in laser cladding and the coating morphology. A central composite experiment design (CCD) with 4 factors and 5 levels was designed to carry out the laser cladding experiment, with the 17-4PH as the substrate and the 15-5PH as the cladding powder. The effects of the process parameters of laser cladding (laser power, powder feeding rate, scanning speed, and overlap rate) on the coating morphology characteristics (dilution rate, heat-affected zone area, and surface flatness) were analyzed. The relative importance of each process parameter was assessed by the random forest model. Based on the assessment results, a WOA-RF prediction model was established, and a multiple regression model and a random forest (RF) model were established for comparative analysis. The analysis reveals that the powder feeding rate exerts the most significant impact on the dilution rate, laser power is the key determinant of the area of the heat-affected zone, and the overlap rate predominantly affects surface flatness of the coating. The evaluation of parameter importance indicates that, under the conditions of this experiment, the overlap rate in the predictive model has a minimal contribution to the area of the heat-affected zone, and the laser power has a minimal contribution to the surface flatness of the coating. The WOA-RF model demonstrates superior performance in predicting morphological features across all evaluated metrics, outperforming both the RF model and the multiple regression model, with higher prediction accuracy. The proposed WOA-RF prediction method effectively captures the nonlinear relationship between process parameters and the morphology of multi-track coatings, which is helpful to determine the optimal process parameters to improve the coating morphology.