摘要
针对选择性激光烧结(selective laser sintering,SLS)成型件精度难以控制以及工艺参数优化实验成本高等问题,提出了一种利用人群搜索算法(seeker optimization algorithm,SOA)优化BP(back propagation)神经网络对SLS成型件精度预测的方法。首先选择激光功率、预热温度、扫描速度、扫描间距以及分层厚度五个工艺参数设计正交试验获取样本数据。然后根据SOA算法特有的利己、利他、预动和不确定推理四大行为确定搜索策略,获取BP神经网络最优权值和阈值。最后采用MATLAB建立优化后的BP神经网络预测模型对样本数据进行预测分析,并与传统BP神经网络和粒子群算法(particle swarm optimization,PSO)优化的BP神经网络预测结果进行对比。结果表明:基于SOA-BP神经网络的预测模型具有较高的预测精度,最大绝对误差仅为0.028,对SLS成型件精度的提高和工艺参数的选取具有指导作用。
For the problems of molding parts accuracy is difficult to control and high cost of parameter optimization experiment in selective laser sintering process,a method of using the seeker optimization algorithm to optimize the BP neural network to predict the accuracy of molding parts was proposed.Firstly,five process parameters of laser power,preheating temperature,scanning speed,scanning distance and layer thickness were selected to design orthogonal experiments for sample data.Then,the search strategy was determined according to the four behaviors of self-interest,altruism,pre-action and uncertain reasoning unique to the SOA algorithm,and the optimal weight and threshold of the BP neural network were obtained.Finally,MATLAB was used to establish an optimized BP neural network prediction model to predict and analyze the sample data with comparing the prediction results with the traditional BP neural network and the BP neural network optimized by particle swarm optimization.The results show that the BP neural network prediction model based on SOA has high prediction accuracy,and the maximum absolute error is only 0.028 which has a guiding effect on the improvement of the accuracy of SLS molded parts and the selection of process parameters.
作者
肖亚宁
郭艳玲
张亚鹏
王扬威
李健
XIAO Ya-ning;GUO Yan-ling;ZHANG Ya-peng;WANG Yang-wei;LI Jian(College of Mechanical and Electrical Engineering, Northeast Forestry University, Harbin 150040, China)
出处
《科学技术与工程》
北大核心
2021年第23期9864-9870,共7页
Science Technology and Engineering
基金
国家自然科学基金(52075090)。
关键词
选择性激光烧结
工艺参数
人群搜索算法
BP神经网络
精度预测
粒子群算法
selective laser sintering
process parameters
seeker optimization algorithm
BP neural network
accuracy prediction
particle swarm optimization
