scholarly journals Hidden Markov Modeling of Human Pivoting

2007 ◽  
Vol 19 (4) ◽  
pp. 444-447 ◽  
Author(s):  
Yusuke Maeda ◽  
◽  
Tatsuya Ushioda ◽  
Keyword(s):  
Multidimensional Scaling ◽  
Hidden Markov Models ◽  
Markov Models ◽  
Hidden Markov ◽  
Minimum Description Length ◽  
Human Movement ◽  
Gaussian Mixture ◽  
Markov Modeling ◽  
Metric Multidimensional Scaling ◽  
Hidden Markov Modeling

In modeling human movement using hidden Markov models (HMM), the “optimal” HMM with an appropriate number of states is determined based on the minimum description length (MDL) criterion. Human pivoting, typifying graspless manipulation, is modeled using Gaussian mixture HMMs. Analyzing the obtained HMMs using metric multidimensional scaling (MDS) showed the features of individual movement. Such dissimilarity analysis can be used to validate models of tacit skills in human manipulation.

Acoustic emission diagnostics of corrosion monitoring in prestressed concrete using hidden Markov and semi-Markov models

2020 ◽  
pp. 147592172097693
Author(s):  
Brennan Dubuc ◽  
Konstantinos Sitaropoulos ◽  
Arvin Ebrahimkhanlou ◽  
Salvatore Salamone
Keyword(s):  
Acoustic Emission ◽  
Corrosion Rate ◽  
Hidden Markov Models ◽  
Prestressed Concrete ◽  
Markov Models ◽  
Hidden Markov ◽  
Accelerated Weathering ◽  
Corrosion Monitoring ◽  
Markov Modeling ◽  
Hidden Markov Modeling

This article presents a framework based on hidden Markov modeling for monitoring corrosion damage in prestressed concrete structures through acoustic emission measurements. Both hidden Markov models and hidden semi-Markov models are investigated to determine the merits of each under typical challenges encountered in real-world monitoring scenarios: an unknown initial corrosion condition and an uncertain corrosion rate. To enable hidden semi-Markov models to perform effectively, they are constructed using extended state hidden Markov models in a manner which is able to easily accommodate these scenarios. Furthermore, a framework is proposed for adopting frequency- and topological-based features of acoustic emission data as indicators of corrosion for the models. Accelerated corrosion tests were carried out on two prestressed concrete specimens; one specimen was used for training the models and the other one for testing. The testing specimen was additionally subjected to accelerated weathering to approximate certain conditions found in the field. A combination of established visual inspection, mass loss measurement, and acoustic emission activity analysis was first used to obtain a benchmark corrosion assessment of the specimens. Frequency- and topological-based acoustic emission features were then used as the basis of the hidden Markov modeling framework. The results demonstrate that both hidden Markov models and hidden semi-Markov models are highly reliable under an unknown initial condition. However, in these tests, the hidden semi-Markov model showed slightly greater reliability against error in the assumed corrosion rate. To improve accuracy in this scenario, a procedure for combining predictions from various candidate corrosion rates was also proposed. The diagnostics obtained from this work may provide useful information for established prognostic models of corrosion mechanisms in prestressed concrete.

Sign in / Sign up

Export Citation Format

Share Document