• Journal of the Korean Institute of Intelligent Systems
• /
• v.11 no.3
• /
• pp.191-199
• /
• 2001

This paper presents a new genetic algorithm for job-shop scheduling problems. When we design a genetic algorithm for difficult ordering problems such as job-shop scheduling problems, it is important to design encoding/crossover that is excellent in characteristic preservation and to maintain a diversity of population. We used Job-based order crossover(JOX). Since the schedules generated by JOX are not always active-schedule, we proposed a method to transform them into active schedulesby using the GT method with c)laracteristic preservation. We introduce strategies for maintaining a diversity of the population by eliminating same individuals in the population. Furthermore, we are not used mutation. Experiments have been done on two examples: Fisher s and Thompson s $lO\timeslO and 20\times5$ benchmark problem.

• Journal of applied mathematics & informatics
• /
• v.11 no.1_2
• /
• pp.215-241
• /
• 2003

In this paper, solute transport in heterogeneous aquifers using a modified Fokker-Planck equation (MFPE) is investigated. This newly developed mathematical model is characterised with a time-, scale-dependent dispersivity. A two-dimensional finite volume quadrilateral mesh method (FVQMM) based on a quadrilateral background interpolation mesh is developed for analysing the model. The FVQMM transforms the coupled non-linear partial differential equations into a system of differential equations, which is solved using backward differentiation formulae of order one through five in order to advance the solution in time. Three examples are presented to demonstrate the model verification and utility. Henry's classic benchmark problem is used to show that the MFPE captures significant features of transport phenomena in heterogeneous porous media including enhanced transport of salt in the upper layer due to its parameters that represent the dependence of transport processes on scale and time. The time and scale effects are investigated. Numerical results are compared with published results on the some problems.

• Structural Engineering and Mechanics
• /
• v.8 no.6
• /
• pp.607-622
• /
• 1999

Being a significant mode of deformation, shear effect in addition to the other modes of stretching and bending have been considered to develop two finite element models for the analysis of beams on elastic foundation. The first beam model is developed utilizing the differential-equation approach; in which the complex variables obtained from the solution of the differential equations are used as interpolation functions for the displacement field in this beam element. A single element is sufficient to exactly represent a continuous part of a beam on Winkler foundation for cases involving end-loadings, thus providing a benchmark solution to validate the other model developed. The second beam model is developed utilizing the hybrid-mixed formulation, i.e., Hellinger-Reissner variational principle; in which both displacement and stress fields for the beam as well as the foundation are approxmated separately in order to eliminate the well-known phenomenon of shear locking, as well as the newly-identified problem of "foundation-locking" that can arise in cases involving foundations with extreme rigidities. This latter model is versatile and indented for utilization in general applications; i.e., for thin-thick beams, general loadings, and a wide variation of the underlying foundation rigidity with respect to beam stiffness. A set of numerical examples are given to demonstrate and assess the performance of the developed beam models in practical applications involving shear deformation effect.

• The Transactions of the Korea Information Processing Society
• /
• v.3 no.3
• /
• pp.564-571
• /
• 1996

The logic synthesis systems for table look up(TLU) type field programmable e gate arrays(FPGAs) have so farstudied mostly the combinational logic problem m. This paper presents for mapping a sequential circuit onto a popular table look up architecture, theXilinx 3090 architecture. In thefirst for solving this problem, combinational and sequential elements which have 6 or7 input combinational and sequential elements which haveless thanor equal to 5 inputs. We heavily use the combinational synthesis techniques tosolve the sequential synthesis problem. Our syntheisis approach is very simple, but its results are reasonable. We compare seveal benchmark Examples with sis-pga(map_together and map_separate) synthesis system and the experimental results show that our synthesis system is 17% betterthan sis-pga sequential synthesis system for TLU PGAs.

• Structural Engineering and Mechanics
• /
• v.70 no.2
• /
• pp.221-231
• /
• 2019

This paper performs for the first time a simultaneous optimization for members sections along with semi-rigid beam-to-column connections for space steel frames with fixed, semi-rigid, and hinged bases using a biogeography-based optimization algorithm (BBO) and a genetic algorithm (GA). Furthermore, a member's sections optimization for a fully fixed space frame is carried out. A real and accurate simulation of semi-rigid connection behavior is considered in this study, where the semi-rigid base connections are simulated using Kanvinde and Grilli (2012) nonlinear model, which considers deformations in different base connection components under the applied loads, while beam-to-column connections are modeled using the familiar Frye and Morris (1975) nonlinear polynomial model. Moreover, the $P-{\Delta}$ effect and geometric nonlinearity are considered. AISC-LRFD (2016) specification constraints of the stress and displacement are considered as well as section size fitting constraints. The optimization is applied to two benchmark space frame examples to inspect the effect of semi-rigidity on frame weight and drift using BBO and GA algorithms.

• Steel and Composite Structures
• /
• v.43 no.5
• /
• pp.603-623
• /
• 2022

This article is dedicated to predict the natural frequencies of joined conical shell structures made of Functionally Graded Material (FGM). The structure includes two conical segments. The equivalent material properties are found by using the rule of mixture based on Voigt model. In addition, three well-known patterns are employed for distribution of material properties throughout the thickness of the structure. The main objective of the present research is to propose a novel exponential pattern and obtain the related equivalent material properties. Furthermore, the Donnell type shell theory is used to obtain the governing equations of motion. Note that these equations are obtained by employing First-order Shear Deformation Theory (FSDT). In order to discretize the governing system of differential equations, well-known and efficient semi-analytical scheme, namely Generalized Differential Quadrature Method (GDQM), is utilized. Different boundary conditions are considered for various types of single and joined conical shell structures. Moreover, an applicable modification is considered for the continuity conditions at intersection position. In the first step, the proposed formulation is verified by solving some well-known benchmark problems. Besides, some new numerical examples are analyzed to show the accuracy and high capability of the suggested technique. Additionally, several geometric and material parameters are studied numerically.

• Geophysics and Geophysical Exploration
• /
• v.26 no.2
• /
• pp.84-93
• /
• 2023

The supervised learning-based deep-learning seismic inversion techniques have demonstrated successful performance in synthetic data examples targeting small-scale areas. The supervised learning-based deep-learning seismic inversion uses time-domain wavefields as input and subsurface velocity models as output. Because the time-domain wavefields contain various types of wave information, the data size is considerably large. Therefore, research applying supervised learning-based deep-learning seismic inversion trained with a significant amount of field-scale data has not yet been conducted. In this study, we predict subsurface velocity models using Laplace-domain wavefields as input instead of time-domain wavefields to apply a supervised learning-based deep-learning seismic inversion technique to field-scale data. Using Laplace-domain wavefields instead of time-domain wavefields significantly reduces the size of the input data, thereby accelerating the neural network training, although the resolution of the results is reduced. Additionally, a large grid interval can be used to efficiently predict the velocity model of the field data size, and the results obtained can be used as the initial model for subsequent inversions. The neural network is trained using only synthetic data by generating a massive synthetic velocity model and Laplace-domain wavefields of the same size as the field-scale data. In addition, we adopt a towed-streamer acquisition geometry to simulate a marine seismic survey. Testing the trained network on numerical examples using the test data and a benchmark model yielded appropriate background velocity models.

• Steel and Composite Structures
• /
• v.34 no.5
• /
• pp.625-641
• /
• 2020

The realistic modeling of the beam-column semi-rigid connection in steel frames attracted the attention of many researchers in the past for the seismic analysis of semi-rigid frames. Comparatively less studies have been made to investigate the behavior of steel frames with semi-rigid connections under different types of earthquake. Herein, the seismic behavior of semi-rigid steel frames is investigated under both far and near-field earthquakes. The semi-rigid connection is modeled by the multilinear plastic link element consisting of rotational springs. The kinematic hysteresis model is used to define the dynamic behavior of the rotational spring, describing the nonlinearity of the semi-rigid connection as defined in SAP2000. The nonlinear time history analysis (NTHA) is performed to obtain response time histories of the frame under scaled earthquakes at three PGA levels denoting the low, medium and high-level earthquakes. The other important parameters varied are the stiffness and strength parameters of the connections, defining the degree of semi-rigidity. For studying the behavior of the semi-rigid frame, a large number of seismic demand parameters are considered. The benchmark for comparison is taken as those of the corresponding rigid frame. Two different frames, namely, a five-story frame and a ten-story frame are considered as the numerical examples. It is shown that semi-rigid frames prove to be effective and beneficial in resisting the seismic forces for near-field earthquakes (PGA ≈ 0.2g), especially in reducing the base shear to a considerable extent for the moderate level of earthquake. Further, the semi-rigid frame with a relatively weaker beam and less connection stiffness may withstand a moderately strong earthquake without having much damage in the beams.

• Journal of the Korea Society for Simulation
• /
• v.15 no.4
• /
• pp.51-58
• /
• 2006

Debugging embedded softwares is very different from debugging general softwares. For examples, debugging embedded software requires more information, such as information on power consumption, information on the distribution of executed instructions, information on the distribution of used registers, and information on the amount of clocks consumed during the execution of a program, that is not needed in debugging general softwares. In this paper, we propose more effective method fer debugging embedded softwares using an instruction set simulator for the microprocessor that is executing embedded softwares. In this research, we develop a debugger based on an instruction set simulator for a domestic embedded microprocessor called SE1608 and we shows an effective debugging method using a MiBench program which is widely used to benchmark embedded softwares. The debugging method proposed in this paper is relatively easy to implement and shows many advantages compared with existing debugging methods.

• KIPS Transactions on Software and Data Engineering
• /
• v.7 no.5
• /
• pp.195-204
• /
• 2018

In this paper, we propose both a new loss function and an online hard example mining scheme for improving the performance of single-stage object detectors which use deep convolutional neural networks. The proposed loss function and the online hard example mining scheme can not only overcome the problem of imbalance between the number of annotated objects and the number of background examples, but also improve the localization accuracy of each object. Therefore, the loss function and the mining scheme can provide intrinsically fast single-stage detectors with detection performance higher than or similar to that of two-stage detectors. In experiments conducted with the PASCAL VOC 2007 benchmark dataset, we show that the proposed loss function and the online hard example mining scheme can improve the performance of single-stage object detectors.