Multilevel classification of milling tool wear with confidence estimation

Fish, Randall; Ostendorf, Mari; Bernard, Gary D.; Castañón, David A.

doi:10.1109/tpami.2003.1159947

Cited by 35 publications

(24 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We use a 2-rate HMM with the short-and long-term cepstral features in the fine and coarse scales, respectively, aimed at capturing noisy/quiet transient regimes due to the built-up edge formation in titanium milling. For both the HMM and multi-rate HMM process models, fullyconnected state transition topologies are chosen based on pilot experiments, the absence of any prior knowledge suggesting a left-to-right topology (unlike steel milling [24]), and the observation that noisy/quiet regimes can occur more than once in a cutting pass. Also for both the HMM and multi-rate HMM process models, the observation distributions are modeled by mixture of diagonal Gaussians, and the number of states and mixture components are determined via cross-validation (cf.…”

Section: B Statistical Modelingmentioning

confidence: 99%

“…As such, the detected wear sequence is not required to be monotonic, and the algorithm can recover from bad decisions given additional evidence. Previous work [49], [24] has found that a generalized linear model (GLM) [37] can further improve the original posterior estimates from Equation 5, by modifying them as follows:…”

Section: Decision Makingmentioning

confidence: 99%

“…Common industrial practice of replacing cutters at regular intervals can be ineffective due to wide variations in a tool's lifetime; yet this practice does not completely prevent costly failures, and leads to unnecessary delays. Therefore, automatic monitoring systems that can reliably operate across a variety of conditions are highly desired, and have been the focus of significant amount of research [13], [45], [30], [50], [24].…”

Section: Introductionmentioning

confidence: 99%

“…Recent work with HMMs (e.g. [41] and [24] for milling, [27] and [19] for drilling, and [52] for turning) have shown the utility of temporal information. For well behaved materials, HMMs may be sufficiently powerful, but for other materials the wear may be more accurately predicted by exploiting temporal information over multiple time scales, not only the information in a single scale.…”

Section: Introductionmentioning

confidence: 99%

“…Direct measurements of the wear on cutting edges are difficult to obtain, and most tool-wear monitoring systems rely on sensory signals measured from the cutting process to predict wear. Tools used in our milling task are relatively small (1/2 in diameter), and cutting forces, current, and power are not very sensitive to track wear on such small size tools [14], [24]. We use acoustic vibrations instead, as they have been successfully used in similar tasks [41], [1], [23], [40].…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Multirate Coupled Hidden Markov Models and Their Application to Machining Tool-Wear Classification

Çetin

Ostendorf

Bernard

2007

IEEE Trans. Signal Process.

View full text Add to dashboard Cite

Abstract-This paper introduces multi-rate coupled hidden Markov models (multi-rate HMMs for short) for multiscale modeling of nonstationary processes, extending traditional HMMs from single to multiple time scales with hierarchical representations of the process state and observations. Scales in the multi-rate HMMs are organized in a coarse-to-fine manner with Markov conditional independence assumptions within and across scales, allowing for a parsimonious representation of both shortand long-term context and temporal dynamics. Efficient inference and parameter estimation algorithms for the multi-rate HMMs are given, which are similar to the analogous algorithms for HMMs. The model is applied to the classification of tool wear in titanium milling, for which acoustic emissions exhibit multiscale dynamics and long-range dependence. Experimental results show that the multi-rate extension outperforms HMMs in terms of both wear prediction accuracy and confidence estimation.Index Terms-Hidden Markov model, multi-rate hidden Markov model, multiscale statistical modeling, confidence, tool wear, tool-wear monitoring, and milling.

show abstract

Section: B Statistical Modelingmentioning

confidence: 99%

Section: Decision Makingmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Multirate Coupled Hidden Markov Models and Their Application to Machining Tool-Wear Classification

Çetin

Ostendorf

Bernard

2007

IEEE Trans. Signal Process.

View full text Add to dashboard Cite

show abstract

Editor's Introduction to Chapter 6

Ovaska¹

2004

Computationally Intelligent Hybrid Systems

View full text Add to dashboard Cite

Fusion of soft-computing (SC) and hard-computing (HC) methodologies is a wellknown approach for practicing engineers. Since the utilization of evolutionary computation, fuzzy logic, and neural networks began to emerge in various applications, also complementary unions of SC and HC have been in common use. Nevertheless, the actual term "Fusion of Soft Computing and Hard Computing" was invented as recently as 1998. But why do we need such a term now if the basic ideas have been in applications use already for more than a decade? We will return to this relevant question at the end of the following discussion. Soft computing is sometimes called-particularly in Japan-"human-like information processing." This is a somewhat misleading description, because the human brain has two specialized hemispheres, the left one (logic, mathematics, analysis, serial processing, etc.) and the right one (creativity, recognition, synthesis, parallel processing, etc.), and they are connected closely to one another through the corpus callosum, which consists of over 200 million nerves [1]. Each hemisphere is continually supporting and complementing the activity of the other. Roger W. Sperry shared the 1981 Nobel Prize in Medicine for his discoveries concerning "the functional specialization of the cerebral hemispheres" [2]. Based on his brilliant work that enhanced our comprehension of the higher functions of human brain, obvious analogies have been established between the left brain and HC, as well as between the right brain and SC. Furthermore, the massive bundle of connecting nerves corresponds to the fusion interface between the two computing paradigms; this crucial communications extension to the above brain analogy is hardly ever mentioned in the literature.

show abstract

Tool Wear Monitoring with Hidden Markov Models

Zhu

2021

Springer Series in Advanced Manufacturing

View full text Add to dashboard Cite

Multilevel classification of milling tool wear with confidence estimation

Cited by 35 publications

References 20 publications

Multirate Coupled Hidden Markov Models and Their Application to Machining Tool-Wear Classification

Multirate Coupled Hidden Markov Models and Their Application to Machining Tool-Wear Classification

Editor's Introduction to Chapter 6

Tool Wear Monitoring with Hidden Markov Models

Contact Info

Product

Resources

About