Scholarlink

<p>Selective laser melting (SLM) enables the fabrication of lightweight, complex components, such as those used in aerospace. The properties of SLM-produced parts strongly depend on the processing parameters. This study systematically investigates the effects of laser power, scanning pitch, and scanning speed on the surface integrity of AlSi10Mg alloy to identify an optimal parameter set. Results show that higher scanning speeds combined with moderate laser power reduce surface roughness and porosity, while also increasing hardness, which is attributed to grain refinement from rapid cooling. For the first time, a backpropagation neural network (BPNN) surrogate model coupled with the non-dominated sorting genetic algorithm III (NSGA-III) was developed to optimize the multi-objective surface properties of SLMed AlSi10Mg. The framework identified an optimized parameter set (laser power = 256.17&#xa0;W, scanning speed = 2060.1&#xa0;mm/s, scanning pitch = 52.49&#xa0;μm), achieving a balanced combination of low surface roughness (6.94&#xa0;μm), high microhardness (104.38&#xa0;HV), and low porosity (1.38%). Compared to previous studies that often focused on single objectives, this integrated approach provides a more effective balance between multi-objective optimization and prediction accuracy. The study validates the feasibility of combining NSGA-III and BPNN for SLM parameter optimization and highlights the potential of SLMed AlSi10Mg for high-performance applications.</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

Optimization of Selective Laser Melting Parameters for Enhanced Surface Integrity of AlSi10Mg Alloy Based on NSGA-III and Backpropagation Neural Network

  • 期刊论文

  • 发表日期：

    2026年3月23日

查看全文
  • Miao-Sen Yang,
  • Yuan Lin,
  • Yi-Xin Liu,
  • Bai-Lin Wu,
  • Hai-Rui Zhang,
  • Xi-Yang Liu,
  • Yue Kang,
  • Bin-Feng Lu,
  • Zhi-Ya Han,
  • Li-Ping Tong,
  • Zhi-Xi Zhang,
  • Yong Guo

摘要

Selective laser melting (SLM) enables the fabrication of lightweight, complex components, such as those used in aerospace. The properties of SLM-produced parts strongly depend on the processing parameters. This study systematically investigates the effects of laser power, scanning pitch, and scanning speed on the surface integrity of AlSi10Mg alloy to identify an optimal parameter set. Results show that higher scanning speeds combined with moderate laser power reduce surface roughness and porosity, while also increasing hardness, which is attributed to grain refinement from rapid cooling. For the first time, a backpropagation neural network (BPNN) surrogate model coupled with the non-dominated sorting genetic algorithm III (NSGA-III) was developed to optimize the multi-objective surface properties of SLMed AlSi10Mg. The framework identified an optimized parameter set (laser power = 256.17 W, scanning speed = 2060.1 mm/s, scanning pitch = 52.49 μm), achieving a balanced combination of low surface roughness (6.94 μm), high microhardness (104.38 HV), and low porosity (1.38%). Compared to previous studies that often focused on single objectives, this integrated approach provides a more effective balance between multi-objective optimization and prediction accuracy. The study validates the feasibility of combining NSGA-III and BPNN for SLM parameter optimization and highlights the potential of SLMed AlSi10Mg for high-performance applications.