基于广义回归神经网络的选区激光烧结成形收缩率研究  

Research on Contraction Rate in SLS Based on Generalized Regression Neural Network

作　　者：刘硕[1] 贺可太[1] 刘锦辉[2] 

机构地区：[1]北京科技大学机械工程学院,北京100083 [2]黑龙江科技大学先进制造研究所,黑龙江哈尔滨150027

出　　处：《铸造技术》2017年第7期1654-1658,共5页Foundry Technology

基　　金：国家自然科学基金资助项目(51274043;71601009)

摘　　要：针对激光烧结增材制造过程中出现的几何尺寸误差问题,采用正交实验与测量的方法获得训练样本,依据广义回归神经网络,建立了选择性激光烧结过程中工艺参数与成形收缩率之间的定量模型,以预测收缩率。定性分析了预热温度与支撑厚度对收缩率的影响,得到了各因素对收缩率影响的权重,并分析了主要因素间的交互作用。通过定性分析与定量预测,可为烧结过程中优化控制收缩提供一个新思路。Using orthogonal experiment and measuring approach to acquire training samples, a quantifiable model between process parameters and modeling contraction rate in selective laser sintering was established based on generalized regression neural network to predict contraction rate. The influence of preheat temperature and supporting thickness on contraction rate was qualitatively analyzed, and the weight of various factors＇ effect on contraction rate was obtained to further analyze the interaction among different factors： By the means of qualitative data analysis and quantifiable prediction, the method can provide a new ideas for optimized controlling contraction rate in sintering process.

关 键 词：激光烧结 增材制造 广义回归神经网络 收缩率 

分 类 号：TB383.3[一般工业技术—材料科学与工程] TN249[电子电信—物理电子学]