HMM with explicit state duration for prognostics in face milling

In this paper, the development of hidden Markov model with explicit state duration (Variable duration HMM) for face milling residual life distribution prognostics is presented. An HMM with explicit state duration is constructed by involving explicit state duration probability. The HMM with explicit...

Full description

Saved in:
Bibliographic Details
Main Authors: Wu Yue, Hong, G S, Wong, Y S
Format: Conference Proceeding
Language:English
Subjects:
Condition monitoring
Error analysis
Exponential distribution
Gaussian distribution
Handwriting recognition
hidden Markov model
Hidden Markov models
Mechanical engineering
Milling
prognostics
Speech recognition
tool condition monitoring
Weibull distribution
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In this paper, the development of hidden Markov model with explicit state duration (Variable duration HMM) for face milling residual life distribution prognostics is presented. An HMM with explicit state duration is constructed by involving explicit state duration probability. The HMM with explicit state duration offers significant advantages over the conventional HMM in prognostics. The reason why including explicit duration has been both verified theoretically and experimentally in this paper. Moreover, two types of state duration pdf (Gaussian and Weibull distribution) have also been studied. VDHMM based prognostics is demonstrated with the case study which is face milling application. In the case study, the mean residual life calculated from both conventional HMM and VDHMM has been compared with the natural mean residual life. The results of the case study has shown that including the state duration as both Gaussian and Weibull distribution perform better than the conventional HMM.
ISSN:2158-2181
DOI:10.1109/RAMECH.2010.5513187