Supervised Gaussian Process Latent Variable Model for Dimensionality Reduction

The Gaussian process latent variable model (GP-LVM) has been identified to be an effective probabilistic approach for dimensionality reduction because it can obtain a low-dimensional manifold of a data set in an unsupervised fashion. Consequently, the GP-LVM is insufficient for supervised learning tasks (e. g., classification and regression) because it ignores the class label information for dimensionality reduction. In this paper, a supervised GP-LVM is developed for supervised learning tasks, and the maximum a posteriori algorithm is introduced to estimate positions of all samples in the latent variable space. We present experimental evidences suggesting that the supervised GP-LVM is able to use the class label information effectively, and thus, it outperforms the GP-LVM and the discriminative extension of the GP-LVM consistently. The comparison with some supervised classification methods, such as Gaussian process classification and support vector machines, is also given to illustrate the advantage of the proposed method.

A meso elastoplastic constitutive model for polycrystalline metals based on equivalent slip systems with latent hardening

A meso material model for polycrystalline metals is proposed, in which the tiny slip systems distributing randomly between crystal slices in micro-grains or on grain boundaries are replaced by macro equivalent slip systems determined by the work-conjugate principle. The elastoplastic constitutive equation of this model is formulated for the active hardening, latent hardening and Bauschinger effect to predict macro elastoplastic stress-strain responses of polycrystalline metals under complex loading conditions. The influence of the material property parameters on size and shape of the subsequent yield surfaces is numerically investigated to demonstrate the fundamental features of the proposed material model. The derived constitutive equation is proved accurate and efficient in numerical analysis. Compared with the self-consistent theories with crystal grains as their basic components, the present theory is much simpler in mathematical treatment.

Time-Frequency Analysis of Nonlinear Systems: The Skeleton Linear Model and the Skeleton Curves

The joint time-frequency analysis method is adopted to study the nonlinear behavior varying with the instantaneous response for a class of S.D.O.F nonlinear system. A time-frequency masking operator, together with the conception of effective time-frequency region of the asymptotic signal are defined here. Based on these mathematical foundations, a so-called skeleton linear model (SLM) is constructed which has similar nonlinear characteristics with the nonlinear system. Two skeleton curves are deduced which can indicate the stiffness and damping in the nonlinear system. The relationship between the SLM and the nonlinear system, both parameters and solutions, is clarified. Based on this work a new identification technique of nonlinear systems using the nonstationary vibration data will be proposed through time-frequency filtering technique and wavelet transform in the following paper.

A constitutive model for cavitation and cavity growth in rubber-like materials under arbitrary tri-axial loading

In this paper, we attempted to construct a constitutive model to deal with the phenomenon of cavitation and cavity growth in a rubber-like material subjected to an arbitrary tri-axial loading. To this end, we considered a spherical elementary representative volume in a general Rivlin's incompressible material containing a central spherical cavity. The kinematics proposed by [Hou, H.S., Abeyaratne, R., 1992. Cavitation in elastic and elastic-plastic solids. J. Mech. Phys. Solids 40, 571-722] was adopted in order to construct an approximate but optimal field. In order to establish a suitable constitutive law for this class of materials, we utilized the homogenisation technique that permits us to calculate the average strain energy density of the volume. The cavity growth was considered through a physically realistic failure criterion. Combination of the constitutive law and the failure criterion enables us to describe correctly the global behaviour and the damage evolution of the material under tri-axial loading. It was shown that the present models can efficiently reproduce different stress states, varying from uniaxial to tri-axial tensions, observed in experimentations. Comparison between predicted results and experimental data proves that the proposed model is accurate and physically reasonable. Another advantage is that the proposed model does not need special identification work, the initial Rivlin's law for the corresponding incompressible material is sufficient to form the new law for the compressible material resulted from cavitation procedure. (C) 2007 Elsevier Ltd. All rights reserved.

Performance of a Supersonic Model Combustor using Vaporized Kerosene Injection

Characteristics of supersonic combustion by injecting kerosene vapor into a Mach 2.5 crossflow at various preheat temperatures and pressures were investigated experimentally. A two-stage heating system has been designed and tested, which can prepare heated kerosene of 0.8 kg up to 820 K at pressure of 5.5 Mpa with minimum/negligible fuel coking. In order to simulate the thermophysical properties of kerosene over a wide range of thermodynamic conditions, a three-component surrogate that matches the compound class of the parent fuel was employed. The flow rate of kerosene vapor was calibrated using a sonic nozzle. Computed flow rates using the surrogate fuel are in agreement with the experimental data. Kerosene jets at various preheat temperatures injecting into both quiescent environment and Mach 2.5 crossflow were visualized. It was found that at injection pressure of 4 Mpa and preheat temperature of 550 K the kerosene jet was completely in vapor phase, while keeping almost the same penetration depth as compared to the liquid kerosene injection. Supersonic combustion tests were also carried out to compare the combustor performance for the cases of vaporized kerosene injection, liquid kerosene injection, and effervescent atomization with hydrogen barbotage, under the similar stagnation conditions. Experimental results demonstrated that the use of vaporized kerosene injection leads to better combustor performance. Further parametric study on vaporized kerosene injection in a supersonic model combustor is needed to assess the combustion efficiency as well as to identify the controlling mechanism for the overall combustion enhancement.

Smoothing LDA Model for Text Categorization

Abstract. Latent Dirichlet Allocation (LDA) is a document level language model. In general, LDA employ the symmetry Dirichlet distribution as prior of the topic-words’ distributions to implement model smoothing. In this paper, we propose a data-driven smoothing strategy in which probability mass is allocated from smoothing-data to latent variables by the intrinsic inference procedure of LDA. In such a way, the arbitrariness of choosing latent variables'priors for the multi-level graphical model is overcome. Following this data-driven strategy,two concrete methods, Laplacian smoothing and Jelinek-Mercer smoothing, are employed to LDA model. Evaluations on different text categorization collections show data-driven smoothing can significantly improve the performance in balanced and unbalanced corpora.

Design Study of a Heavily Loaded Ice Class Propeller Using an Advance Panel Method

A design and optimization procedure developed and used for a propeller installed on a twin-semitunnel-hull ship navigating in very shallow and icy water under heavy load conditions is presented. The base propeller for this vessel was first determined using classic design routines under open-water condition with existing model test data. In the optimization process, a panel method code (PROPELLA) was used to vary the pitch values and distributions and take into account the inflow wake distribution, tunnel gap, and cavitation effects. The optimized propeller was able to improve a ship speed of 0.02 knots higher than the desired speed and 0.06 knots higher than the classic B-series propeller. The analysis of the effect of inflow wake, hull tunnel, cavitation, and blade rake angle on propulsive performance is the focus of this paper.

Study on the two class glass transition and ion transport of a comb polymer electrolyte

The thermal behaviour and ion-transport properties of a comb polymer electrolyte CP350/LiSCN based on methyl vinyl ether/maleic anhydride copolymer with oligo-oxyethylene side chains were studied by means of DSC and ac impedance method. The two glass transition temperatures which can be attributed to side chains and main chains respectively were found to increase with increasing salt concentration. Conductivities which displayed non-Arrhenius behaviour were analyzed by using Vogel-Tammann-Fulcher equation and interpreted on the basis of the configurational entropy model derived by Gibbs and coworkers. The optimum ionic conductivity at 25 degrees C achieved was 2.19x10(-5)S/cm.

Studies on the two class glass transition and the typical VTF behaviour of an amorphous comb polymer electrolyte

A comb polymer (CP350) with oligo-oxyethylene side chains of the type -(CH2CH2O)(7)CH3 was prepared from methyl vinyl ether/maleic anhydride copolymer and poly(ethylene glycol) methyl ether. The polymer can dissolve LiNO3 salt to form homogeneous amorphous polymer electrolyte. This electrolyte system was first found to have two class glass transitions, and the two T(g)s were observed to increase with increasing salt content. The ionic conduction was measured by using the complex impedance method, and conductivities were investigated as functions of temperature and salt concentration. At 25 degrees C, the ionic conductivity maximum of this system can get to 3.72 X 10(-5) S/cm at the [Li]/ [EO] ratio of 0.057. The appearance of the conductivity maximum has been interpreted as being due to the effect of T-g and the so called physical crosslinks. The temperature dependence of the ionic conductivity displaying non-Arrhenius behaviour can be analyzed using the Vogel-Tammann-Fulcher equation and interpreted on the basis of the configurational entropy model.