基于多层感知机模型的熔融沉积尺寸误差预测方法

Dimension Error Predictions in the Fused Deposition Modeling Based on the Multilayer Perceptron Model

熔融沉积成型(FDM)或熔丝制造(FFF),是当下最常见和广泛使用的3D打印技术之一,可用于制造各种功能性构件。然而,FDM(FFF)制件普遍存在尺寸精度低、表面质量差、易翘曲和机械强度不足等现象。目前普遍采用工艺参数优化方法来解决这些问题,但往往需要大量的实验工作和复杂的数据处理。因此,本文以碳纤维增强复合材料的熔融沉积3D打印为例,提出一种基于多层感知机(MLP)模型的FDM(FFF)尺寸误差预测方法。实验结果表明,通过采用4个隐藏层数、神经元节点数常规设计的4-Layers-a网络结构,MLP模型能够实现对尺寸误差的预测,准确率均达到95%以上,可有效应用于FDM(FFF)的工艺参数优化。

Nowadays,the fused deposition modeling(FDM)or fused filament fabrication(FFF)is one of the most common and widely used 3D printing technologies,which could be used to manufacture various functional components.However,part defects of low dimensional accuracy,poor surface quality,susceptibility to warping,and insufficient mechanical strength often occur after printing.These problems are usually solved by optimizing the process parameters,but extensive experiments and complex data processing works are required.Therefore,taking the 3D printing of carbon fiber-reinforced composites as the example,this paper proposed a dimension error prediction method in FDM(FFF)based on the multilayer perceptron(MLP)model.The experimental results indicate that by using the 4-Layers-a network structure with four hidden layers and a conventional design of neuroses,the MLP model could predict the size error.The error prediction accuracy is over 95%,meaning that the 4-Layers-a MLP model is an effective tool to be employed in the process parameter optimizations for FDM(FFF).