Minh Ta

Minimum entropy estimation as a near maximum-likelihood method and its application in system identification with non-Gaussian noise

We derive the minimum entropy estimation (MEE) method from information theory to show the similarity of this method to the maximum likelihood method for the linear regression problem. The result is a nonparametric-based identification technique that can be applied in any case with iid noise that outperforms estimators in this case, including the popular LS method and a recently-developed (and limited) version of the MEE. Performance-wise, the MEE method is comparable to the expectation-maximization (EM) method. Its application to FIR system identification produces a very efficient implementation of this technique.

Stochastic Search Methods to Improve the Convergence of Adaptive Notch Filters

Adaptive Notch Filters (ANFs) are known to have convergence problems due to their non-quadratic error surface. We propose two approaches to improve the convergence of the ANF. The first approach is based on the method of stochastic search. The second approach checks to see whether the estimated signal is correlated to the measurement or is just filtered white noise. The ANF is reinitialized when the estimated signal is filtered white noise (i.e. when the ANF misses the right frequency). Both of these methods show superior convergence comparing to the classical Nehorai ANF.

Adaptive tracking in the time-frequency plane and its application in causal real-time speech analysis

This paper proposes a causal approach to adaptive estimation of time-frequency localized signals using Adaptive Notch Filters (ANF). By adaptively estimating the envelope of each sinusoidal component, it is possible to specify the tracking quality and restart the ANF unit whenever the tracked sinusoid disappears, as well as preventing the ANF from “tracking” non-existent sinusoids (the frequency mis-lock situation). Employing multiple ANFs allows an efficient approach to tracking time-frequency localized signals such as speech.

Robust Notch Filtering by Combining Adaptation in both Time and Frequency

A fully adaptive infinite impulse response (IIR) notch filter has been proposed before with a very good tracking property over a certain frequency range (Torres and DeBrunner, 1998). It is shown that when the frequency to be notched is closed to DC or the Nyquist frequency (fn) the performance of the IIR notch filter is degraded. A finite impulse response (FIR) notch filter with equal number of coefficients exhibits a reversed performance: the tracking performance is worse than the IIR notch filter in those frequencies far from DC and fn but better than the IIR notch filter in those frequencies close to either DC or fn. Hence, a combination of these two filters is proposed in this paper to improve the tracking performance of the notch filter over the total range of frequencies. Computer simulations show the improved performance of the combined adaptive filter. The design is also simple in computation and robust in tracking a time-varying sinusoidal signal.

Adaptive vibration control of a bridge and heavy truck

Bridge vibration control is an important issue whose purpose is to extend the structural service life of steel bridges. Normally, the bridge is modeled as an elastic beam or plate subject to a moving vehicle. However, the moving truck on a bridge is a complicated problem that must still be researched. In this paper, we propose a solution using a time varying model coupled with a black box identification method. This solution does not require an exact physical model. It uses a hybrid multi-channel adaptive vibration control technique that combines feed forward and feedback active control. Our system is called the Intelligent Vehicle/Bridge System (IVBS).

Adaptive Notch Filter with time-frequency tracking of continuously changing frequencies

We propose in this paper a novel modification of the popular Adaptive Notch Filter (ANF) to improve the tracking of time-varying frequencies. Unlike previous algorithms, our new method incorporates a modeling of frequency variation directly into the cost minimization procedure. Our results show a notable improvement in the frequency estimation performance over earlier methods, and comparisons over a few examples show the general effectiveness of our approach.

Application of Minimum Entropy Estimation in Modal Analysis

We propose a modification of the popular eigensystem realization algorithm (ERA) using the minimum entropy estimation (MEE) method. The eigensystem realization algorithm is an LS method that is used to estimate the system matrices in a state-space representation of a physical structure. The MEE is known for its superior variance estimation as compared to the LS method when the measurement noise is unknown and non-Gaussian. It can be used to give a better estimate of the Markov sequence of a dynamical system, which in turn leads to the estimation of the system matrices and the poles using ERA

Embedded Computer/Communication SubSystem for bridge vibration control

We describe an embedded computer/communication sub-system that is used to implement semi-active control. The objective of the control is to reduce the vibrations in a steel-girder bridge and so increase the lifespan of the bridge. An ancillary goal of the control is to increase the dynamic load-carrying capacity of the bridge. We use a wireless-interconnected network of computing nodes, with a remote logging capability for maintenance and performance monitoring. Using this vibration-control system, the life-span of bridges can be extended more than double.

Improving adaptive tracking in time-frequency plane using varying number of Adaptive Notch Filters

Previous work [1] uses Adaptive Notch Filters (ANF) to track time-varying signals in time-frequency plane with a fixed number of ANFs. However, if too few ANFs are used, some importan components may be missed. On the other hand, if too many ANFs are used, noise is also tracked and some artifacts may be introduced. This paper summary proposes an approach to estimate a number of ANFs in order to track a specific time-varying signal.

A robust active noise control algorithm without identifying secondary path

It has been shown that active noise control (ANC) can be done without identifying the secondary path. However, the method given in based on the cumulative error power is ad-hoc, slow, and not general. In this paper, we propose a new algorithm that is more robust to the ANC-without-identifying-secondary-path problem based on a more general secondary-path decomposition and step-size adaptation.