ESTEGHLAL
Year:2005 | Volume:23 | Issue:2|Papers Count:17

Hidden Markov Model is a popular statisical method that is used in continious and discrete speech recognition. The probability density function of observation vectors in each state is estimated with discrete density or continious density modeling. The performance (in correct word recognition rate) of continious density is higher than discrete density HMM, but its computation complexity is very high, especially in very large discrete utterance recognition problems. For real time implementation of very large discrete utterance recognition, we must use discrete density HMM (DDHMM). To increase the performance of DDHMM, one usual solution is fuzzy interpolation. In this study, we present a new method named Gaussian interpolation. We implemented and compared the performance of two types of interpolation methods for 1500 Persian speech command words. Results show that precision and flexibility of Gaussian interpolation is better thanthose of the fuzzy interpolation.

Methods for calculating Available Transfer Capability (ATC) of the transmission systems may be grouped under Static and Dynamic methods. This paper presents a fast dynamic method for ATC calculations, which considers both Transient Stability Limits and Voltage Stability Limits as terminating criteria. A variation of Energy Function Method is used to determine the transient stability limit and the determinant of the Jacobian matrix of the system is used as an index to determine the voltage stability limit. A novel method is used to approximately calculate this determinant. Combining these two methods, an algorithm that calculates ATC, based on both voltage and angle dynamic stability is presented The advantage of this algorithm, besides considering both voltage and angle dynamic stability, is its high speed This-speed of calculation makes the algorithm a perfect candidate to be used in screening contingencies and determining those cases that need to be further analyzed. To demonstrate the validity, efficiency, and the speed of the new method, it is used to calculate ATC for numerical examples with 1, 3, 7 (CIGREE), 10,30 (IEEE) and US (Iowa State) buses.

One of the difficulties with PWM switching converters is high switching loss and electromagnetic interference due to switching at non-zero voltage and current, which limits the operating frequency. In order to reduce the converter volume and weight (by increasing the frequency) and reducing switching losses, zero voltage and current switching methods are recommended. In this paper, four main zero voltage switching (ZVS) methods in full bridge converters are introduced and compared. These four methods are compared on dead times required to obtain the ZVS, load range at ZVS condition, circulating energy in the switch anti parallel diodes during freewheeling periods and voltage oscillations on rectifying diodes. Finally, the results of a 3KW prototype full bridge ZVS converter with a damp circuit for rectifier diodes oscillations are presented and analyzed.

Several groups of methods are used for speaker adaptation in speech recognition. In some techniques, such as MAP estimation, only the models with available training data are updated. Hence, large amounts of training data are required in order to have significant recognition improvements. In some others, such as MLLR, where several general transformations are applied to model clusters, the results are desirable for small training data, but with increase in the amount of training data, the performance improvement saturates. In this paper, a new approach is introduced that makes use of the advantages of the both mentioned techniques to improve the recognition rate. Here, the models with available training data are trained using MAP while for those with insufficient training data, appropriate prior parameters for MAP estimation are found using MLLR. This technique has yielded better performance in comparison to either MAP or MLLR, in a system based on FARSDAT speech corpus.

In this paper the combination of equalization and turbo decoding is studied. In the iterative decoding of a product code in block turbo coding system, the equalization process is performed within the iteration loop. This research aims to investigate the decision feedback equalizer (DFE) incorporated in the iterative decoding. Simulation results show that the more severe channel interference, the more the gain of the combination of DFE and turbo decoder performance compared with the case that using them individually.

In this paper two-dimensional cutting stock problem with demand has been studied. In this problem, cutting of large rectangular sheets in to specific small pieces, should be carried out, hence for the waste will be minimized. Solving this problem is important to decrease waste materials in any industry that requires cutting of sheets. In most of the papers the demand of pieces has not been usually considered. The cutting problems belong to Np-hard problems. So finding a desirable solution in a suitable time is practically impossible and heuristic methods must be used. A meta heuristic algorithm using SA approach is presented. Then attempts have been made to regulate the SAs parameters. Initial solutions are produced with a rule based algorithm and two internal and main SAs are used that lead to better performance of the algorithm. Due to lack of benchmark or test problems, two procedures for generating random problems is presented and are used to study efficiency of the algorithm. For this purpose, problems about 10 to 50 types of pieces with maximum demands of 2400 are generated and solved using the proposed algorithm. The results indicate that the algorithm capable of finding a solution with less than 6% of waste for problems with 30 types of pieces and total demands of 500.

This research was conducted to introduce scientific fundamentals and available methods for wellhead protection area (capture zone) delineation for drinking water wells in the cities. The results of this study could obviate some demands of the national water and wastewater company in quality control of the drinking water resources by delineation and application of the wellhead protection areas. For this purpose, first by using the available literature, criteria and methods of wellhead protection delineation were introduced Then, (1) fixed radius method, (2) simplified variable shape methods, and (3) flow-transport analytical methods implemented in the computer code WHPA were introduced. The applicability of these methods was shown by some sample calculations for Urmia drinking water wells. Samples of the calculated wellhead protection areas for 36 wells in Urmia City were shown using three analytical modules in WHPA. The effects of the hydrogeologic parameters on the wellhead areas were discussed. When reliable hydrogeologic parameters are available in the region where wells are located, the analytical methods and WHPA code produce accurate results for wellhead protection areas.

Flood hydrograph simulation is affected by uncertainty in Rainfall - Runoff (RR) parameters. Uncertainty of RR parameters in Gharasoo catchment, part of the great Karkheh river basin, is evaluated by Monte-Carlo (MC) approach. A conceptual-distributed model, called Monte- Carlo, was used for basin simulation, in which the basin's hydrograph was determined using the superposition of runoff generated by individual cells dividing the catchment in a raster-based discretization. A narrow parameter range was obtained through application of the MC method. Parameter range depended on goodness -of-fit measures. The results of various goodness-of-fit measures are discussed in this paper. The selected goodness-of-fit measures gave high weight to peak discharge to reduce peak discharge error.

Triaxial consolidated drained, unconfined compression, and CBR tests have been conducted in order to study the stress-strain, strength, and volume change characteristics affine sand specimens reinforced by polymeric fibers made from carpet wastes. The variables are aspect ratio (length/width) and weight percentage of the fibers. The results indicate that the peak strength and total volume change of reinforced specimens increase whereas the maximum elastic modulus decreases as the fiber content increases. The rate of increase in the peak strength and total volume change, however, diminishes with increasing the fiber content. The effect of increase in the aspect ratio on results is similar to that of the fiber content.

In this paper, a numerical scheme is proposed for the multi-fluid compressible flows. This method is applied to the problem of underwater explosion. The proposed scheme is basically the extension of Godunov method in gas dynamic problems to the multi-fluid environments and is second-order accurate in space. In this method, also, the problem of artificial mixing of two different phases on Eulerian grids is prevented by a front tracking technique. The numerical results of this study are in very good agreement with previous numerical and experimental results.

In this paper the laminar flow in the rectangular channel bends is simulated using numerical techniques. The turning angle of the channel bend and the area ratio of the channel cross-section are two important parameters to be examined. For flow simulation, the body fitted 3-D continuity and momentum equations are used and a body fitted general purpose code is developed. The existing results of a tied-diriven cavity and the experimental results from a 90 degree square bend were used for code validation. After the code validation, the effect of the area change in the 90 degree bend is examined.The numerical results indicated that increasing the area causes changes in the flow pattern, in turn, which has a direct impact on pressure drop. Similar results were obtained for other bend angles including 30,60,120,150 and 180 degree bends. The results showed that increased bend turning angle increases the pressure drop which is in good agreement with existing experimental data.

On electropolished steel, at low current densities, morphology and texture of zinc electrode posited were investigated. Zinc coating is consisted of hexagonal crystallite laid on each other to produce packets. These packets have different size, and are stacked in different orientations to construct a homogeneous coating on steel substrate. This coating does not have texture, i.e. it has random texture. With increase of current density, the morphology changes completely as each grain attains a special orientation. In this case, coating has a strong basal (0002) along with low angle planes (1013 and 1014). Coating obtained on mechanically polished surfaces is consisted of individual packets of zinc crystals, which are near each other with different orientations. These coatings have a higher density of basal plane (0002) in comparison to electropolished surfaces. The morphology and texture variations with cathodic polarization and surface preparation of steel are due to their effect on nucleation and growth.

In the present research, thermochemical treatment in H2/NH3 atmosphere was used as a post-treatment for electroless Ni-P coatings on the AISI 4140 steel substrates. The high phosphorus (9%) coatings with thicknesses of 2, 24 and 48 µm were applied and the effect of the thermochemical treatment on the morphology, structural changes, roughness, hardness and wear resistance of coatings were studied by SEM, EDS, XRD, profilometry, and microhardness tester. Wear test was used to evaluate wear characteristics of coatings. The wear behaviour of the thermochemical treated/Ni-P coated samples was assessed by comparison with thermochemical treated/uncoated (nitrided) samples. The results showed that effect of thermochemical treatment varies with the coating thickness. In addition, it was shown that a multicomponent coating containing phosphide, nitride and intermetallic phases as well as diffusion region can be developed in the thin (2 µm) electroless Ni-P coated steel by thermochemical treatment. This sample showed better wear resistance than 24 µm Ni-P coated steel under higher load. This behavior was ascribed to nitride phases formed at the surface as well as a nitrogen diffusion zone at the subsurface of thin Ni-P coated steel.

Zinc-Nickel electrodeposits have been widely adopted for surface treatment of automobile steel sheet for high corrosion resistance. In this work the effect of pulse parameters on the Zn-Ni alloy electrode posits were investigated. The hardness, thickness, corrosion resistance and composition of deposits thus produced were investigated. The surface topography of the deposits was also observed in SEM and results are reported. It has been shown that the thickness of the pulse electrode posits was almost even. The hardness in the pulse electrode posits increased by increasing the on-time period and by decreasing the current density. It was also noticed that increasing the on-time period increases the nickel content of the deposit. Pulse electrodeposits had fine structure and the structure become finer by reducing the on-time period and pulse frequency. The corrosion behaviors of the deposits were then investigated. The results showed that the corrosion resistance of the DC electrode posits improves in their nickel content increases. Pulse electrodeposits show the same behavior, but deposits with about 13% nickel show maximum corrosion resistance.

The rapid increase of non-linear loads in the last three decades has caused electrical quantities such as voltages and currents in 3-phase distribution systems to become distorted waveforms. This paper reviews, explains and discusses some new concepts, definitions and new available theory in the unbalanced and distorted systems. The usefulness of the proposed practical definitions is investigated by means of applying them to the real-world measurements of a three-phase electric arc furnace voltages and currents. Finally, the main power terms such as fundamental and non-fundamental power components, harmonic pollution, phase unbalance, active, reactive; nonactive and distortion components are evaluated for a 3-phase arc furnace.