全部 标题 作者
关键词 摘要

OALib Journal期刊
ISSN: 2333-9721
费用:99美元

查看量下载量

相关文章

更多...
-  2015 

不完备先验知识下的刀具磨损状态评估方法研究
Tool Wear Condition Assessment Based on Incomplete Priori Knowledge

Keywords: 刀具状态评估,隐马尔可夫模型,小波包分析,不完备数据
calculations
,efficiency,eigenvalues and eigenfunctions,experiments,feature extraction,feedforward control,hidden Markov models,incomplete knowledge,matrix algebra,measurements,monitoring,probability density function,probability distributions,sensors,tool condition assessment,wavelet packet analysis

Full-Text   Cite this paper   Add to My Lib

Abstract:

在实际切削加工中刀具磨损的全状态先验知识获取困难,而刀具磨钝状态下的先验知识则较易获取。针对这种不完备先验知识情况,以切削力为监测信号,提出基于连续隐马尔可夫模型(CHMM)的刀具磨损状态评估技术。应用小波包分解技术提取信号特征信息,利用刀具磨钝状态下的先验归一化特征信息建立CHMM监测模型;根据刀具未知状态特性向量与监测模型间的对数似然度获取刀具性能指标,实现刀具磨损状态评价。铣刀全寿命磨损实验表明:该方法能在仅具备磨钝状态先验知识情况下,实现对刀具的磨损状态的初步评估,且所需样本数较少,训练速度快。
The priori knowledge of each tool wear state usually can't be completely attained in practical processing, but the priori knowledge of blunt state can be easily attained. In view of this situation, a tool wear monitoring method with incomplete knowledge based on continuous hidden Markov model (CHMM) was proposed in this paper. The cutting force signals were measured as monitoring signals by multi-sensors. The wear features were extracted by wavelet package decomposition technology, and the normalization features of blunt states were inputted to CHMM to construct monitoring models. The tool performance value (PV) can be got through calculating the log-likelihood between unknown state feature vectors and monitoring model, which can achieve the aim of tool condition evaluation. The whole life-cycle wear data of milling cutter were used to validate the effectiveness of the proposed method. The experimental result shows that this method can carry out an accurate assessment of the tool wear when only having the priori knowledge of tool blunt state, and the model has fast learning ability and needs few training samples

Full-Text

Contact Us

service@oalib.com

QQ:3279437679

WhatsApp +8615387084133