A backward-simulation based Rao-Blackwellized particle smoother for conditionally linear Gaussian models

In this article, we develop a new Rao-Blackwellized Monte Carlo smoothing algorithm for conditionally linear Gaussian models. The algorithm is based on the forward-filtering backward-simulation Monte Carlo smoother concept and performs the backward simulation directly in the marginal space of the non-Gaussian state component while treating the linear part analytically. Unlike the previously proposed backward-simulation based Rao-Blackwellized smoothing approaches, it does not require sampling of the Gaussian state component and is also able to overcome certain normalization problems of two-filter smoother based approaches. The performance of the algorithm is illustrated in a simulated application. © 2012 IFAC.

Rao-Blackwellized particle smoothers for mixed linear/nonlinear state-space models

We consider the smoothing problem for a class of conditionally linear Gaussian state-space (CLGSS) models, referred to as mixed linear/nonlinear models. In contrast to the better studied hierarchical CLGSS models, these allow for an intricate cross dependence between the linear and the nonlinear parts of the state vector. We derive a Rao-Blackwellized particle smoother (RBPS) for this model class by exploiting its tractable substructure. The smoother is of the forward filtering/backward simulation type. A key feature of the proposed method is that, unlike existing RBPS for this model class, the linear part of the state vector is marginalized out in both the forward direction and in the backward direction. © 2013 IEEE.

Bayesian inference and learning in Gaussian process state-space models with Particle MCMC

State-space models are successfully used in many areas of science, engineering and economics to model time series and dynamical systems. We present a fully Bayesian approach to inference and learning (i.e. state estimation and system identification) in nonlinear nonparametric state-space models. We place a Gaussian process prior over the state transition dynamics, resulting in a flexible model able to capture complex dynamical phenomena. To enable efficient inference, we marginalize over the transition dynamics function and, instead, infer directly the joint smoothing distribution using specially tailored Particle Markov Chain Monte Carlo samplers. Once a sample from the smoothing distribution is computed, the state transition predictive distribution can be formulated analytically. Our approach preserves the full nonparametric expressivity of the model and can make use of sparse Gaussian processes to greatly reduce computational complexity.

Application of particle swarm optimisation to evaluation of polymer cure kinetics models

A particle swarm optimisation approach is used to determine the accuracy and experimental relevance of six disparate cure kinetics models. The cure processes of two commercially available thermosetting polymer materials utilised in microelectronics manufacturing applications have been studied using a differential scanning calorimetry system. Numerical models have been fitted to the experimental data using a particle swarm optimisation algorithm which enables the ultimate accuracy of each of the models to be determined. The particle swarm optimisation approach to model fitting proves to be relatively rapid and effective in determining the optimal coefficient set for the cure kinetics models. Results indicate that the singlestep autocatalytic model is able to represent the curing process more accurately than more complex model, with ultimate accuracy likely to be limited by inaccuracies in the processing of the experimental data.

Effects of particle plasticity characteristics on local interface stress in particle reinforced composite during uniaxial tension

For elastoplastic particle reinforced metal matrix composites, failure may originate from interface debonding between the particles and the matrix, both elastoplastic and matrix fracture near the interface. To calculate the stress and strain distribution in these regions, a single reinforcing particle axisymmetric unit cell model is used in this article. The nodes at the interface of the particle and the matrix are tied. The development of interfacial decohesion is not modelled. Finite element modelling is used, to reveal the effects of particle strain hardening rate, yield stress and elastic modulus on the interfacial traction vector (or stress vector), interface deformation and the stress distribution within the unit cell, when the composite is under uniaxial tension. The results show that the stress distribution and the interface deformation are sensitive to the strain hardening rate and the yield stress of the particle. With increasing particle strain hardening rate and yield stress, the interfacial traction vector and internal stress distribution vary in larger ranges, the maximum interfacial traction vector and the maximum internal stress both increase, while the interface deformation decreases. In contrast, the particle elastic modulus has little effect on the interfacial traction vector, internal stress and interface deformation.

Micro-mechanical analysis of bonding failure in a particle-filled composite

Composites with a weak interface between the filler and matrix which are susceptible to interfacial crack formation are studied. A finite-element model is developed to predict the stres/strain behavior of particulate composites with an interfacial crack. This condition can be distinguished as a partially bonded inclusion. Another case arises when there is no bonding between the inclusion and the matrix. In this latter case the slip boundary condition is imposed on the section of the interface which remains closed. The states of stress and displacement fields are obtained for both cases. The location of any further deformation through crazing or shear band formation is identified as the crack tip. A completely unbonded inclusion with partial slip at a section of the interface reduces the concentration of the stress at the crack tip. Whereas this might lead to slightly higher strength, it decreases the load-transfer efficiency and stiffness of this type of composite. © 2002 Elsevier Science Ltd. All rights reserved.

Using particle dispersal models to assist in the conservation and recovery of the overexploited native oyster (Ostrea edulis) in an enclosed sea lough

Oyster populations around the world have seen catastrophic decline which has been largely attributed to overexploitation, disease and pollution. While considerable effort and resources have been implemented into restoring these important environmental engineers, the success of oyster populations is often limited by poor understanding of site-specific dispersal patterns of propagules. Water-borne transport is a key factor controlling or regulating the dispersal of the larval stage of benthic marine invertebrates which have limited mobility. The distribution of the native oyster Ostrea edulis in Strangford Lough, Northern Ireland, together with their densities and population structure at subtidal and intertidal sites has been documented at irregular intervals between 1997 and 2013. This paper revisits this historical data and considers whether different prevailing environmental conditions can be used to explain the distribution, densities and population structure of O. edulis in Strangford Lough. The approach adopted involved comparing predictive 2D hydrodynamic models coupled with particle tracking to simulate the dispersal of oyster larvae with historical and recent field records of the distribution of both subtidal and intertidal, populations since 1995. Results from the models support the hypothesis that commercial stocks of O. edulis introduced into Strangford Lough in the 1990s resulted in the re-establishment of wild populations of oysters in the Northern Basin which in turn provided a potential source of propagules for subtidal populations. These results highlight that strategic site selection (while inadvertent in the case of the introduced population in 1995) for the re-introduction of important shellfish species can significantly accelerate their recovery and restoration.

Nonlinear system identification using particle swarm optimisation tuned radial basis function models

A novel particle swarm optimisation (PSO) tuned radial basis function (RBF) network model is proposed for identification of non-linear systems. At each stage of orthogonal forward regression (OFR) model construction process, PSO is adopted to tune one RBF unit's centre vector and diagonal covariance matrix by minimising the leave-one-out (LOO) mean square error (MSE). This PSO aided OFR automatically determines how many tunable RBF nodes are sufficient for modelling. Compared with the-state-of-the-art local regularisation assisted orthogonal least squares algorithm based on the LOO MSE criterion for constructing fixed-node RBF network models, the PSO tuned RBF model construction produces more parsimonious RBF models with better generalisation performance and is often more efficient in model construction. The effectiveness of the proposed PSO aided OFR algorithm for constructing tunable node RBF models is demonstrated using three real data sets.