摘要
针对加工中心刀具剩余寿命预测的实际需求,提出了一种基于信息融合的刀具剩余寿命在线预测方法.通过在加工中心主要功能部件上安装传感器,实时采集加工中心运行过程中的动态信号,经信号预处理和特征参数提取,采用皮尔逊相关系数和残差分析相结合的方法进行特征降维,获得最优的特征参数集.建立基于自适应神经模糊推理系统的刀具剩余寿命预测模型,在线预测刀具剩余寿命.实例分析结果显示:该预测方法的预测结果平均准确率为95.21%,可以满足实际需求.同时,将该预测方法与BP神经网络及其变异模型进行了对比,发现该预测方法预测精度更高.
In order to accurately predict the remaining useful life of cutting tool,a method based on information fusion was proposed in this paper.First,different types of sensors were installed on the principal functional units of machining centers,and sensor signals were collected.Then signal preprocessing and feature extraction were applied.Pearson correlation coefficient and residual error analysis were combined together to implement feature reduction.Afterwards,a remaining useful life prediction model of cutting tool was built based on adaptive-network-based fuzzy inference system,and the obtained optimal features were inputted into the model.Real-time remaining useful life of the cutting tool might be acquired.Finally,a case study was implemented.It is shown that the average accuracy using the proposed method is 95.21%.Meanwhile,compared to back propagation neural network and its variation model using several statistical indices,the proposed method has better prediction performance.
出处
《华中科技大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2017年第4期1-5,共5页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
基金
国家自然科学基金资助项目(51475189
51375181)
中央高校基本科研业务费专项资金资助项目(2016YXMS050)
关键词
加工中心
自适应神经模糊推理系统
剩余寿命预测
信息融合
特征降维
machine center
adaptive-network-based fuzzy inference system
remaining useful life prediction
information fusion
feature reduction
