Conference Paper

Empirical mode decomposition and time varying modelling for carotid audio signal analysis

To read the full-text of this research, you can request a copy directly from the authors.


Auscultation is the acoustic diagnosis of the internal sounds of the body, using a stethoscope. It is performed for examination of the sounds of the circulatory and respiratory systems. Experienced clinicians can hear the flow of blood e.g. in the carotid arteries. We assume that this sound will change over the human life time due to changes inside the vessels. If it is possible to hear the blood flow, it should also be possible to automatically measure changes in the sound of the blood flow. The main problem is that it is necessary to detect dynamical changes at a really long term in a signal that is highly short-term nonstationary. In this work a time-varying (TV) Empirical Mode Decomposition (EMD) analysis was performed to find a trace in the carotid audio signal that is invariable in long-term spaced recordings. EMD has been proposed in the literature as an adaptive time-frequency signal analysis method for dealing with processes involving nonlinear and nonstationary characteristics. Here EMD is used to identify dynamical changes of the carotid blood flow that could be characteristic of each subject through the decomposition of the carotid audio signal in different modes. For that a stethoscope with a microphone were combined with a smartphone for the acquisition of carotid audio signals from five volunteer subjects at different dates in an interval of two months. The recorded signals were first filtered using a wavelet based band-pass filter. The signals were then decomposed using EMD and for some selected modes TV AR models were computed. Finally the TV spectrum and poles were calculated for analysis. Preliminary results show that the TV poles of some modes of the audio signal can be different from subject to subject, and the idea is to further investigate these patterns for patient specific very long-term monitoring.

No full-text available

Request Full-text Paper PDF

To read the full-text of this research,
you can request a copy directly from the authors.

ResearchGate has not been able to resolve any citations for this publication.
Full-text available
A new scheme is proposed that combines autoregressive (AR) modelling techniques and pole-related spectral decomposition for the study of incipient single-point bearing defects for a vibration-based condition monitoring system. Vibration signals obtained from the ball bearings from the high vacuum (HV) and low vacuum (LV) ends of a dry vacuum pump run in normal and faulty conditions are modelled as time-variant AR series. The appearance of spurious peaks in the frequency domain of the vibration signatures translates to the onset of defects in the rolling elements. As the extent of the defects worsens, the amplitudes of the characteristic defect frequencies’ spectral peaks increase. This can be seen as the AR poles moving closer to the unit circle as the severity of the defects increase. The number of poles equals the AR model order. Although not all of the poles are of interest to the user. It is only the poles that have angular frequencies close to the characteristic bearing defect frequencies that are termed the ‘critical poles’ and are tracked for quantification of the main spectral peaks. The time-varying distance, power and frequency components can be monitored by tracking the movement of critical poles. To test the efficacy of the scheme, the proposed method was applied to increasing frame sizes of vibration data captured from a pump in the laboratory. It was found that a sample size of 4000 samples per frame was sufficient for almost perfect detection and classification when the AR poles’ distance from the centre of unit circle was used as the fault indicator. The power of the migratory poles was an alternative perfect classifier, which can be used as a fault indicator. The analysis has been validated with actual data obtained from the pump. The proposed method has interesting potential applications in condition monitoring, diagnostic and prognostic-related systems. Journal of Sound and Vibration 317 (2008) 975–993
In time series modeling, the behavior of a certain phenomenon is expressed in relation to the past values of itself and other covariates. Since many important phenomena in statistical analysis are actually time series and the identification of conditional distribution of the phenomenon is an essential part of the statistical modeling, it is very important and useful to learn fundamental methods of time series modeling. Illustrating how to build models for time series using basic methods, Introduction to Time Series Modeling covers numerous time series models and the various tools for handling them. The book employs the state-space model as a generic tool for time series modeling and presents convenient recursive filtering and smoothing methods, including the Kalman filter, the non-Gaussian filter, and the sequential Monte Carlo filter, for the state-space models. Taking a unified approach to model evaluation based on the entropy maximization principle advocated by Dr. Akaike, the author derives various methods of parameter estimation, such as the least squares method, the maximum likelihood method, recursive estimation for state-space models, and model selection by the Akaike information criterion (AIC). Along with simulation methods, he also covers standard stationary time series models, such as AR and ARMA models, as well as nonstationary time series models, including the locally stationary AR model, the trend model, the seasonal adjustment model, and the time-varying coefficient AR model. With a focus on the description, modeling, prediction, and signal extraction of times series, this book provides basic tools for analyzing time series that arise in real-world problems. It encourages readers to build models for their own real-life problems.
Huang's data-driven technique of Empirical Mode Decomposition (EMD) is presented, and issues re- lated to its effective implementation are discussed. An umber of algorithmic variations, including new stopping criteria and an on-line version of the al- gorithm, are proposed. Numerical simulations are used for empirically assessing performance elements related to tone identification and separation. The obtained results support an interpretation of the method in terms of adaptive constant-Q filter banks.