粘弹性材料钻削温度预测模型的建立与应用
Establishment and Application of Drilling Temperature Prediction Model for Viscoelastic Materials
李冬阳 1张张 2徐志刚 1白鑫林1
作者信息
- 1. 中国科学院沈阳自动化研究所机器人国家重点实验室,辽宁 沈阳 110179;东北大学机械工程与自动化学院,辽宁 沈阳 110819;中国科学院机器人与智能制造创新研究院,辽宁 沈阳 110169
- 2. 上海航天化工应用研究所,浙江 湖州 313000
- 折叠
摘要
粘弹性材料兼有弹性固体和粘性流体的特性,传统的热分析方法无法准确描述同时发生的粘性和弹性变形.针对该复杂材料特性的材料去除过程中的温度场研究是十分稀少的,提出将机器学习算法应用到药柱钻削加工的温度预测当中.首先利用线性回归进行拟合,侧面证明该模型的非线性.其次利用随机森林和BP神经网络算法进行建模,随机森林模型的随机特性使其具有一定的抗噪能力,BP神经网络模型精度最高,决定系数达到0.989.利用BP神经网络模型的预测功能实现加工过程的提前防控,优化加工参数组合,实现高效安全生产.
Abstract
Viscoelastic materials have the properties of both elastic solid and viscous fluid.Traditional thermal analysis methods cannot accurately describe the simultaneous viscous and elastic deformation.The research of temperature field in material remov-al process is very rare according to the characteristics of the complex material.The machine learning algorithm is applied to the temperature prediction of the drilling of the drug string.Firstly,the nonlinear model is proved by linear regression fitting.Second-ly,the random forest and BP neural network algorithm were used for modeling.The random characteristics of the random forest model made it have a certain anti-noise ability,and the BP neural network model had the highest accuracy and the determina-tion coefficient reached 0.989.The prediction function of BP neural network model is used to prevent and control the processing process in advance,optimize the combination of processing parameters,and realize efficient and safe production.
关键词
机器学习/粘弹性材料/钻削温度/预测模型Key words
Machine Learning/Viscoelastic Materials/Drilling Temperature/Prediction Mode引用本文复制引用
基金项目
辽宁省"兴辽英才"计划(XLYC1808040)
出版年
2024
NETL
NSTL
万方数据