基于GWO-DELM算法的20Cr13不锈钢激光熔覆层几何质量特性研究
Geometric Quality Characteristics of Laser Cladding Layer of 20Cr13 Stainless Steel Based on GWO-DELM Algorithm
李永亮 1杜彦斌 1何国华1
作者信息
- 1. 重庆工商大学制造装备机构设计与控制重庆市重点实验室,重庆 400067
- 折叠
摘要
为研究不同工艺参数与激光熔覆层几何质量特性之间的影响关系,建立单道次激光熔覆层几何质量特性:稀释率与宽高比的高精度预测模型.通过建立全因子激光熔覆试验,完成了 20Cr13不锈钢表面熔覆15-5PH合金粉末涂层的几何质量特性数据测量,进行了因子工艺影响分析、回归分析.基于回归模型及预测结果,构建了基于GWO-DELM智能算法的几何质量特性分析模型.结果表明,三元二次回归模型中,稀释率与宽高比的预测精度均极低,预测能力差,而经过GWO-DELM智能算法训练后的模型,预测精度远优于多项式与DELM模型,泛化性较好.基于全因子激光熔覆工艺试验,应用GWO-DELM 智能算法可建立精度较好的单道次熔覆层几何质量特性的分析模型,为复合激光熔覆层几何质量特性控制提供理论支撑.
Abstract
The purpose of this paper is to investigate the influence of different process parameters on the surface geometric quality characteristics of laser cladding layers and to develop a high accuracy prediction model for the geometric characteristics of single pass laser cladding layers.The surface geometric quality characteristics of the laser cladding layer mainly include the dilution rate and the aspect ratio.The paper designed and completed a full factorial laser cladding experiment of 15-5PH alloy powder coating on the surface of 20Cr13 stainless steel.Based on this experiment,data acquisition of geometric quality char-acteristics was completed by applying sample processing,tissue observation,and numerical processing.For the analysis of ex-perimental results,this paper conducted single-factor process influence analysis and ternary quadratic regression analysis.Based on the regression model and prediction results,a prediction model and analysis method of geometric quality characteris-tics based on GWO-DELM intelligent algorithm were constructed.The results of the regression analysis showed that the pre-diction accuracy of both the dilution rate and the aspect ratio were extremely low and the prediction ability was poor.In order to construct a prediction model of geometric quality characteristics with high accuracy,this paper constructed an analytical model based on the GWO-DELM intelligent algorithm,and trains and tests the factorial experimental data.The evaluation in-dex results were much better than the polynomial and DELM regression models,and the prediction accuracy was better.Con-clusion:Based on the experiments of all-factor laser cladding process,the application of GWO-DELM intelligent algorithm can establish the prediction model of surface geometric quality characteristics of single-pass cladding layer with good accuracy.This model and analysis method can provide theoretical support for surface geometric quality control of composite laser clad-ding layers.
关键词
激光熔覆/全因子试验/几何质量特性/灰狼算法/深度极限学习机Key words
laser cladding/full factor test/geometric quality characteristics/grey wolf optimizer/deep extreme learning machine引用本文复制引用
基金项目
重庆市自科基金联合基金重点项目(CSTB2022NSCQ-LZX0011)
重庆市自科面上项目(cstc2020jcyjmsxmX0276)
重庆英才计划(CQYC20210302226)
出版年
2024
NETL
NSTL
万方数据