Statistical Methods for Accelerated Life Testing with Particular Reference to Resin Impregnated Paper Bushings

Electrical failure of insulation is known to be an extremal random process wherein nominally identical pro-rated specimens of equipment insulation, at constant stress fail at inordinately different times even under laboratory test conditions. In order to be able to estimate the life of power equipment, it is necessary to run long duration ageing experiments under accelerated stresses, to acquire and analyze insulation specific failure data. In the present work, Resin Impregnated Paper (RIP) a relatively new insulation system of choice used in transformer bushings, is taken as an example. The failure data has been processed using proven statistical methods, both graphical and analytical. The physical model governing insulation failure at constant accelerated stress has been assumed to be based on temperature dependent inverse power law model.

Efficient Methods for Robust Classification Under Uncertainty in Kernel Matrices

In this paper we study the problem of designing SVM classifiers when the kernel matrix, K, is affected by uncertainty. Specifically K is modeled as a positive affine combination of given positive semi definite kernels, with the coefficients ranging in a norm-bounded uncertainty set. We treat the problem using the Robust Optimization methodology. This reduces the uncertain SVM problem into a deterministic conic quadratic problem which can be solved in principle by a polynomial time Interior Point (IP) algorithm. However, for large-scale classification problems, IP methods become intractable and one has to resort to first-order gradient type methods. The strategy we use here is to reformulate the robust counterpart of the uncertain SVM problem as a saddle point problem and employ a special gradient scheme which works directly on the convex-concave saddle function. The algorithm is a simplified version of a general scheme due to Juditski and Nemirovski (2011). It achieves an O(1/T-2) reduction of the initial error after T iterations. A comprehensive empirical study on both synthetic data and real-world protein structure data sets show that the proposed formulations achieve the desired robustness, and the saddle point based algorithm outperforms the IP method significantly.

Modified Newton, rank-1 Broyden update and rank-2 BFGS update methods in helicopter trim: A comparative study

Nonlinear equations in mathematical physics and engineering are solved by linearizing the equations and forming various iterative procedures, then executing the numerical simulation. For strongly nonlinear problems, the solution obtained in the iterative process can diverge due to numerical instability. As a result, the application of numerical simulation for strongly nonlinear problems is limited. Helicopter aeroelasticity involves the solution of systems of nonlinear equations in a computationally expensive environment. Reliable solution methods which do not need Jacobian calculation at each iteration are needed for this problem. In this paper, a comparative study is done by incorporating different methods for solving the nonlinear equations in helicopter trim. Three different methods based on calculating the Jacobian at the initial guess are investigated. (C) 2011 Elsevier Masson SAS. All rights reserved.

Fast Likelihood Computation in Speech Recognition using Matrices

Acoustic modeling using mixtures of multivariate Gaussians is the prevalent approach for many speech processing problems. Computing likelihoods against a large set of Gaussians is required as a part of many speech processing systems and it is the computationally dominant phase for Large Vocabulary Continuous Speech Recognition (LVCSR) systems. We express the likelihood computation as a multiplication of matrices representing augmented feature vectors and Gaussian parameters. The computational gain of this approach over traditional methods is by exploiting the structure of these matrices and efficient implementation of their multiplication. In particular, we explore direct low-rank approximation of the Gaussian parameter matrix and indirect derivation of low-rank factors of the Gaussian parameter matrix by optimum approximation of the likelihood matrix. We show that both the methods lead to similar speedups but the latter leads to far lesser impact on the recognition accuracy. Experiments on 1,138 work vocabulary RM1 task and 6,224 word vocabulary TIMIT task using Sphinx 3.7 system show that, for a typical case the matrix multiplication based approach leads to overall speedup of 46 % on RM1 task and 115 % for TIMIT task. Our low-rank approximation methods provide a way for trading off recognition accuracy for a further increase in computational performance extending overall speedups up to 61 % for RM1 and 119 % for TIMIT for an increase of word error rate (WER) from 3.2 to 3.5 % for RM1 and for no increase in WER for TIMIT. We also express pairwise Euclidean distance computation phase in Dynamic Time Warping (DTW) in terms of matrix multiplication leading to saving of approximately of computational operations. In our experiments using efficient implementation of matrix multiplication, this leads to a speedup of 5.6 in computing the pairwise Euclidean distances and overall speedup up to 3.25 for DTW.

Review of trend detection methods and their application to detect temperature changes in India

Present study performs the spatial and temporal trend analysis of annual, monthly and seasonal maximum and minimum temperatures (t(max), t(min)) in India. Recent trends in annual, monthly, winter, pre-monsoon, monsoon and post-monsoon extreme temperatures (t(max), t(min)) have been analyzed for three time slots viz. 1901-2003,1948-2003 and 1970-2003. For this purpose, time series of extreme temperatures of India as a whole and seven homogeneous regions, viz. Western Himalaya (WH), Northwest (NW), Northeast (NE), North Central (NC), East coast (EC), West coast (WC) and Interior Peninsula (IP) are considered. Rigorous trend detection analysis has been exercised using variety of non-parametric methods which consider the effect of serial correlation during analysis. During the last three decades minimum temperature trend is present in All India as well as in all temperature homogeneous regions of India either at annual or at any seasonal level (winter, pre-monsoon, monsoon, post-monsoon). Results agree with the earlier observation that the trend in minimum temperature is significant in the last three decades over India (Kothawale et al., 2010). Sequential MK test reveals that most of the trend both in maximum and minimum temperature began after 1970 either in annual or seasonal levels. (C) 2012 Elsevier B.V. All rights reserved.

Assessment of the Relationships Among Design Methods, Design Activities, and Creativity

This paper presents an experimental study that was conducted to compare the results obtained from using different design methods (brainstorming (BR), functional analysis (FA), and SCAMPER) in design processes. The objectives of this work are twofold. The first was to determine whether there are any differences in the length of time devoted to the different types of activities that are carried out in the design process, depending on the method that is employed; in other words, whether the design methods that are used make a difference in the profile of time spent across the design activities. The second objective was to analyze whether there is any kind of relationship between the time spent on design process activities and the degree of creativity in the solutions that are obtained. Creativity evaluation has been done by means of the degree of novelty and the level of resolution of the designed solutions using creative product semantic scale (CPSS) questionnaire. The results show that there are significant differences between the amounts of time devoted to activities related to understanding the problem and the typology of the design method, intuitive or logical, that are used. While the amount of time spent on analyzing the problem is very small in intuitive methods, such as brainstorming and SCAMPER (around 8-9% of the time), with logical methods like functional analysis practically half the time is devoted to analyzing the problem. Also, it has been found that the amount of time spent in each design phase has an influence on the results in terms of creativity, but results are not enough strong to define in which measure are they affected. This paper offers new data and results on the distinct benefits to be obtained from applying design methods. DOI: 10.1115/1.4007362]

Fast acoustic likelihood computation using low-rank matrix approximation

Acoustic modeling using mixtures of multivariate Gaussians is the prevalent approach for many speech processing problems. Computing likelihoods against a large set of Gaussians is required as a part of many speech processing systems and it is the computationally dominant phase for LVCSR systems. We express the likelihood computation as a multiplication of matrices representing augmented feature vectors and Gaussian parameters. The computational gain of this approach over traditional methods is by exploiting the structure of these matrices and efficient implementation of their multiplication.In particular, we explore direct low-rank approximation of the Gaussian parameter matrix and indirect derivation of low-rank factors of the Gaussian parameter matrix by optimum approximation of the likelihood matrix. We show that both the methods lead to similar speedups but the latter leads to far lesser impact on the recognition accuracy. Experiments on a 1138 word vocabulary RM1 task using Sphinx 3.7 system show that, for a typical case the matrix multiplication approach leads to overall speedup of 46%. Both the low-rank approximation methods increase the speedup to around 60%, with the former method increasing the word error rate (WER) from 3.2% to 6.6%, while the latter increases the WER from 3.2% to 3.5%.

Automatic road extraction using high resolution satellite images based on Level Set and Mean Shift methods

Analysis of high resolution satellite images has been an important research topic for urban analysis. One of the important features of urban areas in urban analysis is the automatic road network extraction. Two approaches for road extraction based on Level Set and Mean Shift methods are proposed. From an original image it is difficult and computationally expensive to extract roads due to presences of other road-like features with straight edges. The image is preprocessed to improve the tolerance by reducing the noise (the buildings, parking lots, vegetation regions and other open spaces) and roads are first extracted as elongated regions, nonlinear noise segments are removed using a median filter (based on the fact that road networks constitute large number of small linear structures). Then road extraction is performed using Level Set and Mean Shift method. Finally the accuracy for the road extracted images is evaluated based on quality measures. The 1m resolution IKONOS data has been used for the experiment.

Motion segmentation using curve fitting on Lagrangian particle trajectories

In this paper we present a segmentation algorithm to extract foreground object motion in a moving camera scenario without any preprocessing step such as tracking selected features, video alignment, or foreground segmentation. By viewing it as a curve fitting problem on advected particle trajectories, we use RANSAC to find the polynomial that best fits the camera motion and identify all trajectories that correspond to the camera motion. The remaining trajectories are those due to the foreground motion. By using the superposition principle, we subtract the motion due to camera from foreground trajectories and obtain the true object-induced trajectories. We show that our method performs on par with state-of-the-art technique, with an execution time speed-up of 10x-40x. We compare the results on real-world datasets such as UCF-ARG, UCF Sports and Liris-HARL. We further show that it can be used toper-form video alignment.

Determination of size-independent specific fracture energy of concrete mixes by two methods

The RILEM work-of-fracture method for measuring the specific fracture energy of concrete from notched three-point bend specimens is still the most common method used throughout the world, despite the fact that the specific fracture energy so measured is known to vary with the size and shape of the test specimen. The reasons for this variation have also been known for nearly two decades, and two methods have been proposed in the literature to correct the measured size-dependent specific fracture energy (G(f)) in order to obtain a size-independent value (G(F)). It has also been proved recently, on the basis of a limited set of results on a single concrete mix with a compressive strength of 37 MPa, that when the size-dependent G(f) measured by the RILEM method is corrected following either of these two methods, the resulting specific fracture energy G(F) is very nearly the same and independent of the size of the specimen. In this paper, we will provide further evidence in support of this important conclusion using extensive independent test results of three different concrete mixes ranging in compressive strength from 57 to 122 MPa. (c) 2013 Elsevier Ltd. All rights reserved.

Fastest Rapid Loading Methods of Vertical and Radial Consolidations

Fastest curve-fitting procedures are proposed for vertical and radial consolidations for rapid loading methods. In vertical consolidation, the next load increment can be applied at 50-60% consolidation (or even earlier if the compression index is known). In radial consolidation, the next load increment can be applied at just 10-15% consolidation. The effects of secondary consolidation on the coefficient of consolidation and ultimate settlement are minimized in both cases. A quick procedure is proposed in vertical consolidation that determines how far is calculated from the true , where is coefficient of consolidation. In radial consolidation no such procedure is required because at 10-15% the consolidation effects of secondary consolidation are already less in most inorganic soils. The proposed rapid loading methods can be used when the settlement or time of load increment is not known. The characteristic features of vertical, radial, three-dimensional, and secondary consolidations are given in terms of the rate of settlement. A relationship is proposed between the coefficient of the vertical consolidation, load increment ratio, and compression index. (C) 2013 American Society of Civil Engineers.

Kinetic proofreading at single molecular level: aminoacylation of tRNA(Ile) and the role of water as an editor

Proofreading/editing in protein synthesis is essential for accurate translation of information from the genetic code. In this article we present a theoretical investigation of efficiency of a kinetic proofreading mechanism that employs hydrolysis of the wrong substrate as the discriminatory step in enzyme catalytic reactions. We consider aminoacylation of tRNA(Ile) which is a crucial step in protein synthesis and for which experimental results are now available. We present an augmented kinetic scheme and then employ methods of stochastic simulation algorithm to obtain time dependent concentrations of different substances involved in the reaction and their rates of formation. We obtain the rates of product formation and ATP hydrolysis for both correct and wrong substrates (isoleucine and valine in our case, respectively), in single molecular enzyme as well as ensemble enzyme kinetics. The present theoretical scheme correctly reproduces (i) the amplitude of the discrimination factor in the overall rates between isoleucine and valine which is obtained as (1.8x10(2)).(4.33x10(2)) = 7.8x10(4), (ii) the rates of ATP hydrolysis for both Ile and Val at different substrate concentrations in the aminoacylation of tRNA(Ile). The present study shows a non-michaelis type dependence of rate of reaction on tRNA(Ile) concentration in case of valine. The overall editing in steady state is found to be independent of amino acid concentration. Interestingly, the computed ATP hydrolysis rate for valine at high substrate concentration is same as the rate of formation of Ile-tRNA(Ile) whereas at intermediate substrate concentration the ATP hydrolysis rate is relatively low. We find that the presence of additional editing domain in class I editing enzyme makes the kinetic proofreading more efficient through enhanced hydrolysis of wrong product at the editing CP1 domain.

Capturability of augmented proportional navigation (APN) guidance with nonlinear engagement dynamics

Proportional Navigation (PN) and its variants are widely used guidance philosophies. However, in the presence of target maneuver, PN guidance law is effective only for a restrictive set of initial geometries. To account for target maneuvers, the concept of Augmented Proportional Navigation (APN) guidance law was introduced and analyzed in a linearized interceptor-target engagement framework presented in literature. However, there is no work in the literature, that addresses the capturability performance of the APN guidance law in a nonlinear engagement framework. This paper presents such an analysis and obtains the conditions for capturability. It also shows that a shorter time of interception is obtained when APN is formulated in the nonlinear framework as proposed in this paper. Simulation results are given to support the theoretical findings.

Exploring novel methods to achieve sensitivity limits for high operating temperature infrared detectors

A review of high operating temperature (HOT) infrared (IR) photon detector technology vis-a-vis material requirements, device design and state of the art achieved is presented in this article. The HOT photon detector concept offers the promise of operation at temperatures above 120 K to near room temperature. Advantages are reduction in system size, weight, cost and increase in system reliability. A theoretical study of the thermal generation-recombination (g-r) processes such as Auger and defect related Shockley Read Hall (SRH) recombination responsible for increasing dark current in HgCdTe detectors is presented. Results of theoretical analysis are used to evaluate performance of long wavelength (LW) and mid wavelength (MW) IR detectors at high operating temperatures. (C) 2013 Elsevier B.V. All rights reserved.