直写成形制备FGMs零件时延迟信息的数字化预测方法  

Digital prediction method for delay information for preparing FGMs parts by direct write forming

作　　者：王世杰 王龙 马硕 杨杰[1] 马聪 段国林[1] WANG Shijie;WANG Long;MA Shuo;YANG Jie;MA Cong;DUAN Guolin(College of Mechanical Engineering,Hebei University of Technology,Tianjin 300401,China)

机构地区：[1]河北工业大学机械工程学院,天津300401

出　　处：《工程设计学报》2023年第2期127-135,共9页Chinese Journal of Engineering Design

基　　金：中央引导地方科技发展资金项目(216Z1804G)。

摘　　要：采用直写成形工艺制备FGMs(functionally graded materials,功能梯度材料)零件时存在材料配比变化延迟的现象,导致所制备零件的材料配比与设计目标的吻合度较差,材料性能的未知性增大,从而存在潜在的使用风险。为了准确获取在不同工艺参数下材料配比的延迟信息,构建了一种基于计算流体力学的神经网络预测模型。基于RNG k-ε模型,采用优化后的贝叶斯正则化神经网络模型来预测不同工艺参数所对应的延迟信息,即在不同的材料初始配比、目标配比、螺杆转速、双挤出柱塞进给速率和下预测交付延迟时间和整体延迟时间,其预测精度分别可以达到94.87%与92.74%。采用数字图像处理方法对在不同工艺参数下打印的FGMs试样进行处理,结果表明实际打印试样的材料梯度变化曲线与仿真结果有较高的吻合度,验证了以计算流体力学为分析框架的仿真结果的准确性以及所构建的优化神经网络模型对延迟信息预测的可行性与可靠性。研究结果为未来将数字化预测方法融入FGMs零件本体制备的工艺提供了参考,可以促进传统制造模式向数字化制造模式转变,最终实现FGMs零件的精准制造。When using direct ink writing technology to prepare FGMs(functionally graded materials)parts,there is a delay in the change of material ratio,which leads to poor consistency between the material ratio of the prepared parts and the design goals,and increases the uncertainty of material properties,resulting in potential use risks.In order to accurately learn the delay information of material ratio under different process parameters,a neural network prediction model based on computational fluid dynamics was constructed.Based on RNG k-εmodel,an optimized Bayesian regularized neural network model was used to predict the delay information corresponding to different process parameters,namely,the delivery delay time and global delay time were predicted under different initial ratio and target ratio of materials,screw rotation speed,sum of double extruded plunger feed rates.The prediction accuracy could reach 94.87%and 92.74%,respectively.Using digital image processing methods to process FGMs samples printed under different process parameters,the results showed that the material gradient change curve of the actual printed samples had a high degree of coincidence with the simulation results,verifying the accuracy of the simulation results based on computational fluid dynamics as the analysis framework,as well as the feasibility and reliability of the constructed optimized neural network model for predicting delay information.The research results provide a reference for integrating digital prediction methods into FGMs part body preparation processes in the future,and can promote the transformation of traditional manufacturing models to digital manufacturing models,ultimately achieving the accurate manufacturing of FGMs parts.

关 键 词：功能梯度材料 延迟信息 优化神经网络 数字化预测 

分 类 号：TB34[一般工业技术—材料科学与工程]