• Journal of Korea Multimedia Society
• /
• v.20 no.8
• /
• pp.1406-1420
• /
• 2017

Artificial neural networks (ANNs) play an important role in the fields of function approximation, prediction, and classification. ANN performance is critically dependent on the input parameters, including the number of neurons in each layer, and the optimal values of weights and biases assigned to each neuron. In this study, we apply the particle swarm optimization method, a popular optimization algorithm for determining the optimal values of weights and biases for every neuron in different layers of the ANN. Several regression models, including general linear regression, Fourier regression, smoothing spline, and polynomial regression, are conducted to evaluate the proposed method's prediction power compared to multiple linear regression (MLR) methods. In addition, residual analysis is conducted to evaluate the optimized ANN accuracy for both training and test datasets. The experimental results demonstrate that the proposed method can effectively determine optimal values for neuron weights and biases, and high accuracy results are obtained for prediction applications. Evaluations of the proposed method reveal that it can be used for prediction and estimation purposes, with a high accuracy ratio, and the designed model provides a reliable technique for optimization. The simulation results show that the optimized ANN exhibits superior performance to MLR for prediction purposes.

• Journal of the Society of Naval Architects of Korea
• /
• v.54 no.3
• /
• pp.250-257
• /
• 2017

It is desirable to have a way to predict the pressure drag due to various appendages attached to stern. As a mathematical model for these, a sphere and a singularity behind it, both in the uniform flow can be considered. We may use the Butler's sphere theorem to find the Stokes' stream function when the resulting flow is axisymmetric, and then the extended Lagally's theorem to get the force upon the sphere due to the singularity. Assuming the separation distance between the sphere and the singularity is small, the leading order approximation for the force is obtained and it is found out that if the separation distance and the square root of the strength of the dipole are of the same order, the effect of the image of the dipole with respect to the sphere is the most important.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.9 no.2
• /
• pp.583-599
• /
• 2015

We apply the Bayesian matching pursuit (BMP) algorithm to the estimation of time-frequency selective channels in orthogonal frequency division multiplexing (OFDM) systems. By exploiting prior statistics and sparse characteristics of propagation channels, the Bayesian method provides a more accurate and efficient detection of the channel status information (CSI) than do conventional sparse channel estimation methods that are based on compressive sensing (CS) technologies. Using a reasonable approximation of the system model and a skillfully designed pilot arrangement, the proposed estimation scheme is able to address the Doppler-induced inter-carrier interference (ICI) with a relatively low complexity. Moreover, to further reduce the computational cost of the channel estimation, we make some modifications to the BMP algorithm. The modified algorithm can make good use of the group-sparse structure of doubly selective channels and thus reconstruct the CSI more efficiently than does the original BMP algorithm, which treats the sparse signals in the conventional manner and ignores the specific structure of their sparsity patterns. Numerical results demonstrate that the proposed Bayesian estimation has a good performance over rapidly time-varying channels.

• Journal of international Conference on Electrical Machines and Systems
• /
• v.3 no.4
• /
• pp.428-432
• /
• 2014

Rotating machines are an integral part of large electrical power machinery in most of the industries. Any degradation or outages in the rotating electric machinery can result in significant losses in productivity. It is critical to monitor the equipment for any degradation's so that it can serve as an early warning for adequate maintenance activities and repair. Prior research and field studies have indicated that the rotating machines have a particular type of signal structure during the initial start-up transient. A machine performance can be studied based on the effect of degradation in signal parameters. In this paper a data-acquisition system and the FFT algorithm has been design and model using the MATLAB and Simulink. The implementation had been carried out on the TMS320 DSP Processor and various testing and verification of the machine performance had been carried out. The results show good agreement with expected results for both simulated and real-time data. The real-time data from AC water pumps which have rotating motors built-in were collected and analysed. The FFT algorithm provides frequency response and based on this frequency response performance of the machine had been measured.The FFT algorithm provides only approximation about the machine performances.

• Transactions of the Korean Society of Mechanical Engineers B
• /
• v.20 no.8
• /
• pp.2626-2636
• /
• 1996

A study of soot deposition and reentrainment was carried out both theoretically and experimentally to understand behavior of soot formed by incomplete combustion in a diesel engine. Theoretically, soot deposition on engine cylinder wall and/or piston head was studied with a stagnation point flow approximation. Soot reentrainment occurred upon exhaust gas blowdown was also studied by assuming a long-normal shear velocity distribution. Experimentally, a LPG$O_2/N_2$ flame impinging on a disk, produced by a concentric tubular burner, was chosen as deposition configuration and a shear flow unit with compressed air was installed for the study of reentrainment. For selected flame configuration, soot deposition measurements were conducted and showed that the dominant deposition mechanism was thermophoresis. Distributions of gas temperature and soot number density were estimated by combining data obtained by a B-type thermocouple with a thermophoretic transport theory. Disk temperature distributions were directly measured using a K-type thermocouple. Soot size and morphology were estimated from a TEM photograph. Ratios of soot deposit to reentrained amount were measured for a wide range of shear flow velocities, which showed that the reentrainment model was reasonable.

• Proceedings of the KSME Conference
• /
• 2001.06e
• /
• pp.410-415
• /
• 2001

An analytical and numerical examination of second-order fractional-step methods and boundary condition for the incompressible Navier-Stokes equations is presented. In this study, the compatibility condition for pressure Poisson equation and its boundary conditions, stability, and numerical accuracy of canonical fractional-step methods has been investigated. It has been found that satisfaction of compatibility condition depends on tentative velocity and pressure boundary condition, and that the compatible boundary conditions for type D method and approximately compatible boundary conditions for type P method are proper for divergence-free velocity for type D and approximately divergence-free for type P method. Instability of canonical fractional-step methods is induced by approximation of implicit viscous term with explicit terms, and the stability criteria have been founded with simple model problems and numerical experiments of cavity flow and Taylor vortex flow. The numerical accuracy of canonical fractional-step methods with its consistent boundary conditions shows second-order accuracy except $D_{MM}$ condition, which make approximately first-order accuracy due to weak coupling of boundary conditions.

• Transactions of the Korean Society of Mechanical Engineers
• /
• v.19 no.12
• /
• pp.3329-3343
• /
• 1995

The interaction of mixed convection and surface radiation in a printed circuit board(PCB) is investigated numerically. The electronic equipment is modeled by a two-dimensional channel with three hot blocks. In order to calculate the turbulent flow characteristics, the low Reynolds number k-.epsilon. model which is proposed by Launder and Sharma is applied. The S-4 approximation is used to solve the radiative transfer equation. The effects of the Reynolds number and geometric configuration variation of PCB on the flow and heat transfer characteristics are analyzed. As the results of this study, it is found that the thermal boundary layer occured at adiabatic wall in case with thermal radiation included, and the effect of radiation is also found to be insignificant for high Reynolds numbers. It is found, as well, that the heat transfer increases as the Reynolds number and block space increase and the channel height decreases and the heat transfer of vertical channel is greater than that of horizontal channel.

• Journal of the Institute of Electronics Engineers of Korea SC
• /
• v.42 no.5 s.305
• /
• pp.13-18
• /
• 2005

Over the last few years, the control of bipedal robot has been considered a promising but difficult research field in the community of robotics. In this paper, a new robust output control method for a bipedal robot is proposed using the adaptive fuzzy logic. The adaptive fuzzy logic is used as an system approximator to cancel the unknown uncertainty. First, a model for a bipedal robot including switching leg influence, uncertainty and disturbance is presented. Second, a controller is designed in which the joint velocity measurement is not required. Fuzzy approximation error estimator is inserted in the system for tuning the fuzzy logic. Finally, the result of the computer simulation is presented to show the validity of the suggested control method.

• ETRI Journal
• /
• v.40 no.4
• /
• pp.421-434
• /
• 2018

The application of deep neural networks (DNNs) to connect the world with cyber physical systems (CPSs) has attracted much attention. However, DNNs require a large amount of memory and computational cost, which hinders their use in the relatively low-end smart devices that are widely used in CPSs. In this paper, we aim to determine whether DNNs can be efficiently deployed and operated in low-end smart devices. To do this, we develop a method to reduce the memory requirement of DNNs and increase the inference speed, while maintaining the performance (for example, accuracy) close to the original level. The parameters of DNNs are decomposed using a hybrid of canonical polyadic-singular value decomposition, approximated using a tensor power method, and fine-tuned by performing iterative one-shot hybrid fine-tuning to recover from a decreased accuracy. In this study, we evaluate our method on frequently used networks. We also present results from extensive experiments on the effects of several fine-tuning methods, the importance of iterative fine-tuning, and decomposition techniques. We demonstrate the effectiveness of the proposed method by deploying compressed networks in smartphones.

• Journal of Mechanical Science and Technology
• /
• v.14 no.9
• /
• pp.985-996
• /
• 2000

The issue of quench is related to safety operation of large-scale superconducting magnet system fabricated by cable-in-conduit conductor. A numerical method is presented to simulate the thermal hydraulic quench characteristics in the superconducting Tokamak magnet system, One-dimensional fluid dynamic equations for supercritical helium and the equation of heat conduction for the conduit are used to describe the thermal hydraulic characteristics in the cable-in-conduit conductor. The high heat transfer approximation between supercritical helium and superconducting strands is taken into account due to strong heating induced flow of supercritical helium. The fully implicit time integration of upwind scheme for finite volume method is utilized to discretize the equations on the staggered mesh. The scheme of a new adaptive mesh is proposed for the moving boundary problem and the time term is discretized by the-implicit scheme. It remarkably reduces the CPU time by local linearization of coefficient and the compressible storage of the large sparse matrix of discretized equations. The discretized equations are solved by the IMSL. The numerical implement is discussed in detail. The validation of this method is demonstrated by comparison of the numerical results with those of the SARUMAN and the QUENCHER and experimental measurements.