FDM表面粗糙度偏最小二乘回归建模研究

产品的表面质量差严重制约了熔融沉积成型(fused deposition modeling，FDM)工艺的发展，通过粗糙度成型机理构建参数模型是提升FDM 成型件表面质量的重要途径。文章以某FDM 打印设备成型件的水平表面粗糙度为例，通过探讨参数三次曲线与多段直线拟合的丝宽模型，利用粗糙度的几何机理提出了基于3 个主要工艺参数的回归模型，采用样条变换的非线性偏最小二乘回归(partial least squares，PLSR)建立对所选参数的有效预测模型，并对建模中未覆盖的不同打印参数下的实验数据样本进行校验，分析工艺参数对水平表面粗糙度的影响。通过以上研究方法，提高了水平表面粗糙度预测效率，获取了相关的经验知识。

Research on surface roughness modeling for FDM part based on 〖JZ〗partial least squares regression

The low surface quality of products seriously affects the development of the fused deposition modeling (FDM) process. It is a very effective way to improve the surface roughness of FDM molded parts by constructing a parameter model through the roughness forming mechanism. The roughness of the horizontal surface in some FDM printing device was taken as an example. Based on the geometric mechanism of roughness, a regression model based on three main process parameters was proposed by researching the profile of extruded filament section shape fitted by the parameter cubic curve and the multi-segment straight line. An effective prediction model for selected parameters was established by the method of nonlinear partial least squares regression (PLSR) based on spline transformation afterwards. In order to verify the model, the experimental data which were not included in the samples of the regression model under different process parameters were tested. The influence of process parameters on the horizontal surface roughness was analyzed. Thus, the efficiency of predicting horizontal surface roughness was improved and relevant experience and knowledge was obtained.