Search

Copyright of Journal of Dalian University of Technology / Dalian Ligong Daxue Xuebao is the property of Journal of Dalian University of Technology and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

In the article, several concise and efficient iterated posterior linearization filtering and smoothing methodologies are proposed for nonlinear systems with cross-correlated noises. Based on the Gaussian approximation (GA), the presented methods are derived via performing statistical linear regressions (SLRs) of the nonlinear state-space models w.r.t the current posterior distribution in an iterated way. Various posterior linearization methods can be developed by employing different approximation computation approaches for the Gaussian-weighted integrals encountered in SLRs. These new estimation methods enjoy not only the accuracy and robustness of the GA filter but also the lower computational complexity. Estimation performances of the designed methods are illustrated and compared with conventional estimation schemes by two common numerical examples. • Several methods are derived for the nonlinear system with cross-correlated noises. • The new methods feature concise derivation, higher accuracy and stronger robustness. • An effective balance between estimation accuracy and computational cost is achieved. [ABSTRACT FROM AUTHOR]

The ultimate goal of an artificial pancreas (AP) is finding the optimal insulin rates that can effectively reduce high blood glucose (BG) levels in type 1 diabetic patients. To achieve this, most autonomous closed-loop strategies continuously compute the optimal insulin bolus to be administrated on the basis of the estimated plasma concentrations for glucose and insulin. Unlike subcutaneous glucose levels which can be measured in real-time, unavailability of insulin sensors makes it essential the use of mathematical models so as to fully estimate plasma insulin concentrations. For model-based estimation, GP-Bayesian filters have been recently proposed to incorporate probabilistic non-parametric Gaussian process (GP) models of dynamic systems into Kalman filtering techniques. As a result, model uncertainty can explicitly be incorporated into the prediction step and in the filtering processes, which is usually not the case for more traditional filtering strategies that resort to parametric models for state estimation. More specifically, the question arises as to whether glycemic variability is properly taken into account in model formulations and whether it would compromise proper estimation of plasma insulin concentration. To tackle this, a stochastic glycemic model including variability was incorporated into different parametric and nonparametric filtering techniques to provide an estimate of the plasma insulin levels. In particular, we compared density representation against using knowledge about the parameterization of the transition dynamics and the observation function. We found that, as glycemic variability increases, filtering techniques based on parametric models rapidly degrades their performance as a consequence of large nonlinearities. Results show that Bayes’ filtering techniques increase predictability of the patient state, and thus, boost safety and performance in the AP control and monitoring tasks. [ABSTRACT FROM AUTHOR]

We develop a structural model where joint dynamics of aggregate consumption and asset-specific dividends are governed by correlated state variables. The correlation structure implies distinct cross-sectional exposures of dividends to a long history of consumption growth rates, resulting in variation of consumption beta. Such variation rationalizes momentum crashes per Daniel and Moskowitz (2016), as the consumption beta of the Winner portfolios remain low after the economy recovers from a downturn, while the consumption beta of the Loser portfolios grow quickly. Thus, emerging from a recession, the consumption beta of the momentum strategy decreases, and so does risk premia. [ABSTRACT FROM AUTHOR]

Collaboration between low-quality sensor and high-quality sensor can achieve tradeoff between accuracy and energy efficiency in heterogeneous sensor networks (HSNs). Generally, HSNs are deeply integrated with dynamic physical environments. The monitored target's dynamics are the most important and common factors of the dynamic environments. Some important parameters (such as active opportunity, sampling frequency and sampling time) fail to adapt to the changes, which undermines the collaboration's performance when the state of the monitored target changes. Even the system performance is not up to user requirements or large amounts of energy are consumed. To solve this problem, we propose an adaptive collaboration scheme named EasiAC by the collaboration between magnetic and camera sensors. First, for the dynamics of the monitored target, EasiAC utilizes the magnetic sensors to predict the target's state via Bayesian filtering based method. Second, to achieve good performance of such collaboration, EasiAC adjusts the camera sensors' active opportunity, optimal sampling frequency and sampling time dynamically according to the estimated results from the magnetic sensors. Finally, a boosting based algorithm named BbTC is proposed to make classification for the target to achieve high accuracy (average classification accuracy is more than 98%). We evaluate EasiAC method through extensive simulations and real road environment experiments. The results demonstrate that EasiAC needs less energy consumption (saving 97% energy) than traditional solutions, while maintaining the performance at acceptable level (average image integration ratio is 90%) in the presence of target's dynamics. [ABSTRACT FROM AUTHOR]

To deal with non-linear, non-Gaussian state estimation problem, a kind of particle filter algorithm based on interval estimation was proposed. This paper analyzed the auxiliary particle filter at first. After interval estimating the expectation of the system states, the new algorithm sampled uniformly in the interval and updated the filter results using the new measurement. The interval extension conditions were proposed to ensure the validity of the estimated range and computational efficiency of the algorithm. Sampling uniformly in the interval avoids the particle degeneracy and improves the particle divergence. The experimental results show that the new particle filter is significantly better than the general particle filter. [ABSTRACT FROM AUTHOR]

Abstract: In this paper we describe a stochastic dynamic data-driven application system (DDDAS) for monitoring, in real-time, material damage in aerospace structures. The work involves experiments, different candidate damage models, finite element discretization, Bayesian analysis of the candidate models, Bayesian filtering with the most plausible model, parallel scientific libraries, and high performance computing. Here we describe a low-degree-of-freedom model designed for proof-of-concept, in preparation for the development of the full DDDAS. The physical system involves fiber-reinforced composite plates subjected to quasi-static loading and enriched with distributed carbon nanotubes that act as sensors, signaling damage through changes on the voltage profile. We give an overview of the experimental data we collected, of the damage models we explored, and of the Bayesian methodology we applied in order to use uncertain experimental data for driving the stochastic system. [Copyright &y& Elsevier]

This paper presents a class of signal processing techniques for collaborative signal processing in ad hoc sensor networks, focusing on a vehicle tracking application. In particular, we study two types of commonly used sensors—acoustic-amplitude sensors for target distance estimation and direction-of-arrival sensors for bearing estimation-and investigate how networks of such sensors can collaborate to extract useful information with minimal resource usage. The information-driven sensor collaboration has several advantages: tracking is distributed, and the network is energy-efficient, activated only on a when-needed basis. We demonstrate the effectiveness of the approach to target tracking using both simulation and field data. [ABSTRACT FROM AUTHOR]

[Display omitted] • An adaptive Bayesian framework is proposed for MR image restoration. • Gaussian bilateral range and domain filters as prior are explored for restoration. • Rician likelihood & noise adaptive prior help in edge preservation & bias removal. • Convergence of the proposed filter combination further demonstrate an optimal filter performance • Experiments on Brainweb & clinical database demonstrate encouraging results. MR images contain noise and for quantitative clinical diagnosis restoration is essential. For single coil MR images, noise follows Rician distribution when signal to noise ratio (SNR) is low and Gaussian distribution when SNR is high. Rician noise is signal dependent and introduces bias. The work proposes an adaptive Bayesian framework for restoration of 2D magnitude MR images. Restoration is achieved by Rician likelihood as the data attachment term with range and domain Gaussian filters, adaptive to noise as prior in Maximum A ´ posterior framework. A good filtering behavior is achieved due to the domain component of the filter and crisp edges are preserved at the same time due to the noise adaptive range component. Rician likelihood aids the image restoration in terms of bias removal. Convergence of the proposed method further highlights the optimal filtering performance. Experiments conducted on publically available Brainweb phantom demonstrate enhanced performance in terms of signal to noise ratio, structural similarity index and overall performance. [ABSTRACT FROM AUTHOR]

The accuracy of cooperative localization for multiple autonomous underwater vehicles (AUVs) equipped with low precise proprioceptive localization sensors can be improved by using relative location information between individuals and Bayesian filtering. However, when the relative location measurement errors are high, its accuracy will be reduced. Two measurement for rough estimation algorithms under the constraint environment of the cooperative structure are developed in this paper: the first algorithm is based on the underwater acoustic isotropic transmission. And the second algorithm is based on the common observation environment. In the first algorithm, it builds under the assumption that the distance errors calculated from the simultaneous omnidirectional response signals from the same transmitting source have correlated. Similarly, in the second algorithm, the assumption that "common observation environment" is correlated is made. First, the correlation between the errors is used roughly to estimate the measurement of information. Then, a suitable filter is applied to fuse the rough estimation measurement with dead-reckoning estimation that improves the location estimation accuracy. The final simulation, by changing the AUV formation navigation paths and the sensor observation noises, shows the proposed processing methods have effectiveness and consistency compared to the traditional algorithm. • Two measurement rough estimation algorithms are developed. • One is based on the underwater acoustic isotropic transmission. • The other is based on the common observation environment. • Relevant properties are used to convey information and reduce common mode errors. • The accuracy of co-location is improved by preprocessing the measurement errors. [ABSTRACT FROM AUTHOR]