收藏我们   ENGLISH  
客服QQ
  点击这里给我发消息505255386
微信
  微信公共号
关键词查询
范围:  
提 交
《工程塑料应用》 2024年第01期
DOI:10.3969/j.issn.1001-3539.2024.01.014
基于回归分析和GA-BP神经网络算法的3D打印件弯曲性能预测
白鹤1,杨鑫1,杨瑞琦1,刘亚明23,赵峥璇1,庞瑞4,何石磊23
1.宝鸡职业技术学院机电信息学院,陕西宝鸡 721013; 2.宝鸡石油钢管有限责任公司钢管研究院,陕西宝鸡 721008;  3.国家石油天然气管材工程技术研究中心,陕西宝鸡 721008; 4.宝鸡石油钢管厂职工子弟学校,陕西宝鸡 721008 
Flexural property prediction of 3D-printing sample based on regression analysis and GA-BP neural network
BAI He1, YANG Xin1, YANG Ruiqi1, LIU Yaming23, ZHAO Zhengxuan1, PANG Rui4, HE Shilei23
1.College of Mechatronics and Informatics,Baoji Vocational & Technical College,Baoji 721013,China; 2.Steel Pipe Research Institute of Baoji Petroleum Steel Pipe Co.,Ltd.,Baoji 721008,China; 3.National Petroleum and Gas Tubular Goods Engineering Technology Research Center,Baoji 721008,China; 4.Baoji Steel Pipe Factory School,Baoji 721008,China 
摘要:为进一步探究熔融沉积成型(FDM)3D打印参数和制件弯曲性能之间的关系,创建合理的 FDM 3D 打印制件弯曲强度预测模型。根据正交试验 L16(45)的设计原则和神经网络算法模型的构建要求,按照不同分层高度、填充密度、打印温度 、打印速度以及外壳厚度五种因素,制备25组试验试样,并进行弯曲性能检测。随后通过建立 GA-BP神经网络模型、传统BP神经网络模型以及多元回归方程模型,分别对 FDM 3D 打印制件弯曲性能进行预测,并将预测数据与试验测试数据进行对比。通过对比发现,GA-BP神经网络模型预测数据与试验测试数据更为接近,其平均误差为3.71%,且误差值整体波动最小,BP神经网络模型与多元回归方程模型预测精度相差不大,BP神经网络模型预测平均误差为8.05%,多元回归方程模型预测平均误差为 9.07%,但多元回归方程误差值整体波动最大。因此,采用GA遗传算法优化后的BP神经网络模型在进行FDM 3D打印制件弯曲性能预测方面具有更高的精度和更良好的稳定性。 
Abstract:In order to further explore the relationship between fused deposition modeling(FDM) 3D printing parameters and the flexural property of parts, a reasonable prediction model for the flexural property of FDM 3D printed parts was created. According to the principles of orthogonal experiment L16(45) and the requirements of neural network algorithm model, 25 sets of test specimens were prepared in terms of five factors such as different layer thickness, filling density, nozzle temperature, filling speed and shell thickness. Flexura property was tested. Subsequently, the flexura property of FDM 3D printing parts was predicted by establish ing the GA-BP neural network model, traditional BP neural network model and multiple regression equation model. The predicted data and the experimental test data were compared. It was found by comparison that the predicted data of the GA-BP neural network model was closer to the experimental test data, with an average error of 3.71%. The overall fluctuation of the error is minimal. The prediction accuracy of the BP neural network model and the multiple regression equation model is not significantly different, and the average prediction error of the BP neural network model is 8.05%. The average prediction error of the multiple regression equation model is 9.07%. But the overall fluctuation of the error in the multiple regression equation is the largest. Therefore, the BP neural network model optimized by GA genetic algorithm has higher accuracy and better stability in predicting the flexura property of FDM 3D printed parts. 
关键词:回归分析;GA-BP神经网络;3D打印;弯曲性能;预测 
Keywords:regression analysis; GA-BP neural network; 3D printing; flexural property; prediction 
基金:宝鸡职业技术学院 2023 年度院级课题项目(2023048Z) 
本文引用格式:
白鹤,杨鑫,杨瑞琦,等.基于回归分析和GA-BP神经网络算法的3D打印件弯曲性能预测[J].工程塑料应用,2024,52(1):89-94.
BAI He,YANG Xin,YANG Ruiqi,et al. Flexural property prediction of 3D-printing sample based on regression analysis and GA-BP neuralnetwork[J]. Engineering Plastics Application,2024,52(1):89-94. 

原文:
 
相似文章:
  支具大流量3D打印工艺参数对挤料力的影响
2023年第12期
  纤维缠绕复合材料压力容器的多尺度分析与性能预测
2023年第11期
  连续与短切CF增强PA6复合材料双喷头3D打印工艺参数优化
2023年第11期
  基于SLA成型的薄板件制备与工艺参数优化
2023年第11期
  机器学习在复合材料领域中的应用进展
2023年第09期
  3D打印TPU/PCL共混物食管支架在食管内的生物力学性能
2023年第09期
  固化深度对光固化3D打印成型性能和精度的影响
2023年第09期
  基于辊压增强的连续纤维复材3D打印方法研究
2023年第08期
  光固化3D打印制备PLA/β-TCP生物支架及性能
2023年第08期
  PA66注塑件组装过程开裂的分析及措施
2023年第06期
《工程塑料应用》杂志社 版权所有,未经授权禁止复制或者建立镜像
地址:山东省济南市天桥区田庄东路3号 邮编:250031
电话:0531-85878057 85878223 85878148 85878278
鲁ICP备05018524号-5