Showing papers by "PSG College of Technology published in 2002"
TL;DR: In this paper, X-ray diffraction and scanning electron microscopic studies were carried out to confirm the polycrystalline nature of the chalcopyrite structure of the indium diselenide thin films of different thicknesses.
Abstract: Copper indium diselenide thin films of different thicknesses were prepared by the chemical bath deposition technique onto well-cleaned glass substrates at room temperature. The thickness of the deposited films has been determined by gravimetry. The structural characterization was carried out by X-ray diffraction and scanning electron microscopic studies. These studies confirm the polycrystalline nature of the films with chalcopyrite structure. The structural parameters such as lattice constants, axial ratio, tetragonal distortion, crystallite size, dislocation density and number of crystallites per unit area have been evaluated. The composition of the various constituents in CuInSe 2 films has been determined by energy dispersive X-ray analysis. The optical properties have been studied in detail in the wavelength range 4000-14500 A and the optical band gap has been found to be direct and allowed.
78 citations
TL;DR: In this article, a generative computer-aided process planning (CAPP) system for prismatic components is presented, named PSG-CAPP, which is designed and developed using Visual Basic 6.0 as front end and Oracle 7.3 as back end.
Abstract: This paper deals with design and development of a generative computer-aided process planning (CAPP) system for prismatic components and named PSG-CAPP. The proposed CAPP system has been divided into three modules. The first module is concerned with feature extraction. The second and third modules deal with planning the set-up, machine selection, cutting tool selection, cutting parameter selection, and generation of process plan sheet. The whole CAPP system is developed using Visual Basic 6.0 as front end and Oracle 7.3 as back end. The SolidWorks98 plus has been used for modelling, and feature extraction has been implemented using a program written in Visual Basic 6.0. The speciality of the developed CAPP system is that it is linked with the CAD module and it extracts the majority of features automatically prior to process planning. The developed CAPP system has been tested with aerospace components. The main feature of the proposed CAPP system is its ability to handle a variety of prismatic components and generate process plans for them. A case study has been included to highlight the potential of the CAPP system.
53 citations
TL;DR: In this paper, the results of an extensive study on the microstructure, microhardness, corrosion, and tensile properties of continuously borided and interrupted boriding specimens of medium-carbon steel are compared.
Abstract: The results of an extensive study on the microstructure, microhardness, corrosion, and tensile properties of continuously borided and interrupted borided specimens of medium-carbon steel are compared. Carbon repartitioning away from the surface is one of the principal modes to accommodate the high strains introduced on boron diffusion into the case. However, this is a kinetically constrained process and is more predominant on interrupted boriding. The effect of such a carbon redistribution is to result in microstructural modifications including (1) blunting of boride needle tips, (2) precipitation of nearly spherical and fine borocarbides, and (3) enhanced carbon segregation at the boride needle/steel matrix interface on interrupted boriding. The mechanisms aiding the change in the morphology of the boride needles are discussed. The improvements in the mechanical and corrosion properties of the interrupted borided specimens over continuously borided specimens are described.
41 citations
TL;DR: A sequential orthogonal approach to the building and training of a neural network with a single radial basis function (RBF) neuron is presented in this paper and one hidden RBF neuron itself is sufficient to achieve desired accuracy.
41 citations
TL;DR: An algorithm for computing reliability of K -out-of- N system is given that is an easy to implement, fast and memory efficient algorithm and helps to improve the computational efficiency considerably.
35 citations
TL;DR: In this article, a thermal mass bypass flow meter is described, with details of its design, principle of operation, calibration, and testing of effects of ambient temperature and orientation on its performance.
27 citations
01 Feb 2002
TL;DR: In this article, a finite element analysis (FEA) model was used to predict the effect of fiber percentage on the thrust force causing delamination in bidirectional glass fiber reinforced plastic (GFRP) composite laminates.
Abstract: Drilling experiments were conducted on bidirectional glass fibre-reinforced plastic (GFRP) composite laminates using high-speed steel drill tools in order to verify the effect of machining parameters and fibre concentration on the surface finish, hole quality and delamination. The experimental results revealed that the surface roughness (SR) value and hole diameter variation can be correlated with the drill speed, feed rates and fibre content. A finite element analysis (FEA) model is used to predict the effect of fibre percentage on the thrust force causing delamination. A modified simple plate bending equation derived from linear elastic fracture mechanics (LEFM) is used to validate the FEA results. The predicted FEA results are shown to agree reasonably well with LEFM model results.
23 citations
TL;DR: The Sequential Learning Neural Network (SLNN) model proposed by Zhang and Morris is used in this paper to tackle the common problem encountered by the conventional Feed Forward Neural Network in determining the network structure in the number of hidden layers and theNumber of hidden neurons in each layer.
Abstract: A sequential orthogonal approach to the building and training of a single hidden layer neural network is presented in this paper. The Sequential Learning Neural Network (SLNN) model proposed by Zhang and Morris [1]is used in this paper to tackle the common problem encountered by the conventional Feed Forward Neural Network (FFNN) in determining the network structure in the number of hidden layers and the number of hidden neurons in each layer. The procedure starts with a single hidden neuron and sequentially increases in the number of hidden neurons until the model error is sufficiently small. The classical Gram–Schmidt orthogonalization method is used at each step to form a set of orthogonal bases for the space spanned by output vectors of the hidden neurons. In this approach it is possible to determine the necessary number of hidden neurons required. However, for the problems investigated in this paper, one hidden neuron itself is sufficient to achieve the desired accuracy. The neural network architecture has been trained and tested on two practical civil engineering problems – soil classification, and the prediction o strength and workability of high performance concrete.
14 citations
TL;DR: In this article, the efficacy of octylamine as a vapour phase corrosion inhibitor in different corrosive environments for carbon steel, copper and brass, and its biocidal action, have been examined.
Abstract: The efficacy of octylamine as a vapour phase corrosion inhibitor in different corrosive environments for carbon steel, copper and brass, and its biocidal action, have been examined. The protection of these metals from atmospheric corrosion using VPI (octylamine)‐impregnated paper was monitored by visual observation, weight loss measurements, polarisation scans and impedance measurements. The long‐term protection given by octylamine‐impregnated paper to these metals was examined by Stevenson’s chamber test. It was found that octylamine offered more than 80 per cent protection against atmospheric corrosion for carbon steel, copper and brass.
11 citations
TL;DR: In this paper, a finite-element model of the workpiece to be machined and its subsequent frequency response analysis was used to investigate certain aspects of high-speed machining for improving the accuracy of thin-walled aerospace components.
Abstract: "The study is aimed to investigate certain aspects of high-speed machining for improving the accuracy of thin-walled aerospace components. The approach used involved the development of finite-element model of the workpiece to be machined and its subsequent frequency response analysis. The response of the workpiece subjected to dynamic cutting force gives an indication of the best possible speeds from the point of view of accuracy. Based on the results of the analysis, it is possible to predict the range of spindle speeds at which the workpiece demonstrates very high dynamic rigidity. In addition, this study has established the superiority of high-speed machining to produce aerospace structures with high stiffness-to-weight ratio and also throws some light on the capability of high speed in machining of low rigidity sculptured-surface components. "
6 citations
18 Nov 2002
TL;DR: The potential of Simulated Annealing behaving as a Pattern Recognizer to recognize images even when they appear transformed, viz., rotated, translated and scaled, besides noisy, has been studied.
Abstract: The application of Simulated Annealing for the identification of Military Targets, images of which can be both noisy and noise free, nominal and transformed, has been investigated in this paper. The images of Military Targets of an enemy are captured by an unmanned reconnaissance spy plane and then are interpreted by an expert. In the absence of an expert, Simulated Annealing can be used to identify the enemy targets. The potential of Simulated Annealing behaving as a Pattern Recognizer to recognize images even when they appear transformed, viz., rotated, translated and scaled, besides noisy, has been studied. Performance of Simulated Annealing during different temperatures has also been studied on experimental cases.
TL;DR: A sequential orthogonal approach to the building and training of a single hidden layer fuzzy neural network is presented and a simple defuzzification process is proposed.
25 Nov 2002
TL;DR: The proposed algorithm provides a solution for the blind source separation problem by shifting the domain of the problem to the time-frequency domain and applying ICA to each of the frequency components individually.
Abstract: Blind source separation of acoustic mixtures aims at providing a solution to the classical cocktail-party problem. The inherent delays and convolutions in microphone recordings, entails a modification in the independent component analysis (ICA), which achieves separation by making the assumption of statistical independence of source signals that are linearly combined. The proposed algorithm provides a solution for the blind source separation problem by shifting the domain of the problem to the time-frequency domain and applying ICA to each of the frequency components individually. Satisfactory results were achieved for speech-music as well as speech-speech separation by adopting the time-frequency domain ICA.
07 Aug 2002
TL;DR: A flow sensor based on the principle of constant temperature hot-wire anemometry has been developed to measure the airflow through the mouth of the subject during expiration.
Abstract: Measurement of expiratory parameters provide useful information about overall respiratory system that can be fundamental in categorizing and staging pulmonary diseases. In contrast to the conventional stand-alone instrument, a PC based virtual instrument for the measurement of expiratory parameters has been proposed in this paper. A flow sensor based on the principle of constant temperature hot-wire anemometry has been developed to measure the airflow through the mouth of the subject during expiration. Output signal of the anemometer is connected to a PC through an analog to digital interface card, PCI 6025E from National Instruments. A virtual instrument with all controls, indicators and display required for the measurement and interpretation of expiratory parameters has been developed using the software package LabVIEW.
TL;DR: Two neural dynamic models which belong to the class of recurrent neural networks (RNN) have been formulated for the solution of equilibrium and eigenvalue problems.
06 May 2002
TL;DR: Back propagation neural networks ~BPN! for prediction of shear strength of reinforced-concrete ~RC! deep beams from the experimental data and instead of the BPN one can apply the sequential learning neural network ~SLNN! proposed by Zhang and Morris ~1998! with a single hidden neuron.
Abstract: The writers are to be congratulated for applying back propagation neural networks ~BPN! for prediction of shear strength of reinforced-concrete ~RC! deep beams from the experimental data. In their work the BPN consists of an input layer of ten nodes ~with one bias node!, two hidden layers with six and five nodes, and an output layer of one node. They also used a linear activation function for the input layer and a sigmoidal function for the other layers. They concluded that the learning rate of 0.4 and momentum factor of 0.2 provide the best combinations. V calculated by BPN was not tabulated but graphically presented. The writers would like to point out that instead of the BPN one can apply the sequential learning neural network ~SLNN! proposed by Zhang and Morris ~1998! with a single hidden neuron. This avoids finding the neural network architecture in the number of hidden layers and neurons. The network consists of ten inputs ~for nine inputs1one bias node! and one hidden layer with one hidden neuron and one output neuron denoting the shear strength of RC beams, as shown in Fig. 1. The procedure starts with a single hidden neuron and sequentially increases in the number of hidden neurons until the model error is sufficiently small. We use linear activation functions for input and output neurons and the sigmoidal function for the hidden layer. The classical Gram‐Schmidt orthogonalization method is used at each step to form a set of orthogonal bases spanned by output vectors of the hidden neurons. In this approach it is possible to determine the number of hidden neurons required. However, for the problem investigated, one hidden neuron itself is sufficient to achieve accuracy. For the algorithm, the reader may refer to Zhang and Morris ~1998!. First, as given by the writers, the inputs and output are normalized as
TL;DR: In this paper, a fuzzy logic-based precompensator is developed to linearise the control power plant so that the controller design becomes valid, which is used for a secondary injection thrust vector control system used for the solid booster stage of a launch vehicle.
Abstract: Control forces are required for steering a launch vehicle to guide it to follow an optimal trajectory. Launch vehicle control involves two control loops, the inner loop deals with short-period dynamics, stability and the outer loop, known as the guidance loop, optimises the trajectory. The general nonlinear plant model is first approximated as a linear time-varying plant over a nominal trajectory and then segmented as linear, time-invariant plant models at different time intervals. A major part of the plant model is the control power plant, which for a secondary injection thrust vector control system used for the solid booster stage of a launch vehicle is nonlinear due to various reasons. The controllers designed for different time regimes assume the control power plant as linear and are adapted smoothly by a technique called gain scheduling to cope with the plant model changes wrt time. In this paper, a fuzzy logic-based pre-compensator is developed to linearise the control power plant so that the controller design becomes valid. Simulation results are presented to validate the design and a novel preprocessing technique is developed to reduce the size of the fuzzy inference system.
TL;DR: A novel technique combining both adaptive noise cancellation and adaptive signal enhancement in a single recurrent neural network is proposed to alleviate the computational burden imposed by RTRL algorithm by employing pruning of weights.
Abstract: The electrogastrogram measured cutaneously by attaching electrodes to the abdominal skin contains considerable noise. An attempt is made to develop an efficient adaptive filtering technique suitable for real time processing of electrogastrogram. A novel technique combining both adaptive noise cancellation and adaptive signal enhancement in a single recurrent neural network is proposed. To compare its performance, adaptive noise cancellation and cascaded connection of adaptive noise cancellation and adaptive signal enhancement are employed. Recurrent neural networks using Real Time Recurrent Learning (RTRL) algorithm are employed for implementing the all the above systems. An attempt is made to alleviate the computational burden imposed by RTRL algorithm by employing pruning of weights.
Journal Article•
TL;DR: The proposed IDL compiler is based on the insight that IDLs are true languages amenable to modern compilation techniques, and through the support of intermediate language representation called as Abstract Object Interface (AOI), the compiler can support multiple IDL's and target languages.
Abstract: An interface definition language (IDL) is a traditional language describing the interfaces between the components. IDL compilers generate stubs that provide communicating processes with the abstraction of local object invocation or procedure call. Typical IDL compilers are limited to a single IDL and target language, but the proposed IDL compiler is based on the insight that IDLs are true languages amenable to modern compilation techniques. Through the support of intermediate language representation called as Abstract Object Interface (AOI), our compiler can support multiple IDLs and target languages. Given an IDL (for example CORBA) the IDL compiler can generate stubs and skeletons for different target languages like Java and C++ and for different distributed object technologies like Remote Method Invocation/ Java Remote Method Protocol (RMI/JRMP), RMI/Internet Interoperable Protocol (RMI/IIOP) and Common Object Request Architecture (CORBA/IIOP). Further, interoperability can be achieved between them using a single compiler.