基于BP神经网络的SLS烧结件质量的预测  

The Quality Prediction of SLS Part Based on BP Neural Network

作　　者：任继文[1] 殷金菊[1] 董连杰[1] 

机构地区：[1]华东交通大学机电工程学院,江西南昌330013

出　　处：《机床与液压》2012年第21期15-18,共4页Machine Tool & Hydraulics

基　　金：国家自然科学基金资助项目(51162008);江西省自然科学基金资助项目(20114BAB206001)

摘　　要：选择性激光烧结的烧结件质量预测是一个多变量、非线性的问题,采用传统的方法很难得到满意的结果。采用BP神经网络模型,在数值模拟取样的基础上,建立了烧结件质量的神经网络预测模型。该模型确定了工艺参数激光功率P、扫描速度v和预热温度T0与烧结宽度和烧结深度的关系。其预测结果与数值模拟结果相一致,说明该神经网络模型能定量地反映出工艺参数与烧结件质量之间的关系,据此可合理选择加工工艺参数。The quality prediction of sintered parts was a multi-variable and nonlinear problem in selective laser sintering （ SLS）, which was difficult to obtain satisfactory results by the traditional methods. By adopting BP neural network model, a neural network prediction model for the quality of SLS part was established based on the sampling of numerical simulation, which was used to deter- mine the relation between processing parameters including the laser power, scanning speed and preheating temperature, and quality at- tributes including sintering width and sintering depth. The prediction results are in consistent with the numerical simulation result, which show that the neural network model might be used to reflect the relationship quantitatively between process parameters and quality of SLS parts. So it is a valuable guide to select appropriate process parameters.

关 键 词：选择性激光烧结 BP神经网络 工艺参数 预测 

分 类 号：TN249[电子电信—物理电子学] TG156.99[金属学及工艺—热处理]