• Journal of Institute of Control, Robotics and Systems
• /
• v.10 no.12
• /
• pp.1295-1304
• /
• 2004

In this paper, the Bayesian recurrent neural network is proposed to predict time series data. A neural network predictor requests proper learning strategy to adjust the network weights, and one needs to prepare for non-linear and non-stationary evolution of network weights. The Bayesian neural network in this paper estimates not the single set of weights but the probability distributions of weights. In other words, the weights vector is set as a state vector of state space method, and its probability distributions are estimated in accordance with the particle filtering process. This approach makes it possible to obtain more exact estimation of the weights. In the aspect of network architecture, it is known that the recurrent feedback structure is superior to the feedforward structure for the problem of time series prediction. Therefore, the recurrent neural network with Bayesian inference, what we call Bayesian recurrent neural network (BRNN), is expected to show higher performance than the normal neural network. To verify the proposed method, the time series data are numerically generated and various kinds of neural network predictor are applied on it in order to be compared. As a result, feedback structure and Bayesian learning are better than feedforward structure and backpropagation learning, respectively. Consequently, it is verified that the Bayesian reccurent neural network shows better a prediction result than the common Bayesian neural network.

• Journal of the Korean Society of Combustion
• /
• v.19 no.2
• /
• pp.8-14
• /
• 2014

Experiments were conducted to clarify role of the outermost edge flame on low-strain-rate flame extinction in buoyancy-suppressed non-premixed methane flames diluted with He and $N_2$. The use of He curtain flow produced a microgravity level of $10^{-2}-10^{-3}g$ in $N_2$- and He-diluted non-premixed counterflow flame experiments. The critical He and $N_2$ mole fractions at extinction with a global strain rate were examined at various burner diameters (10, 20, and 25 mm). The results showed that the extinction curves differed appreciably with burner diameter. Before the turning point along the extinction curve, low-strain-rate flames were extinguished via shrinkage of the outermost edge flame with and without self-excitation. High-strain-rate flames were extinguished via a flame hole while the outermost edge flame was stationary. These characteristics could be identified by the behavior of the outermost edge flame. The results also showed that the outermost edge flame was not influenced by radiative heat loss but by convective heat addition and conductive heat losses to the ambient He curtain flow. The numerical results were discussed in detail. The self-excitation before the extinction of a low-strain-rate flame was well described by a dependency of the Strouhal number on global strain rate and normalized nozzle exit velocity.

• Wind and Structures
• /
• v.19 no.3
• /
• pp.339-352
• /
• 2014

This paper discusses the appropriate duration for basic gust wind speeds in wind loading codes and standards, and in wind engineering generally. Although various proposed definitions are discussed, the 'moving average' gust duration has been widely accepted internationally. The commonly-specified gust duration of 3-seconds, however, is shown to have a significant effect on the high-frequency end of the spectrum of turbulence, and may not be ideally suited for wind engineering purposes. The effective gust durations measured by commonly-used anemometer types are discussed; these are typically considerably shorter than the 'standard' duration of 3 seconds. Using stationary random process theory, the paper gives expected peak factors, $g_u$, as a function of the non-dimensional parameter ($T/{\tau}$), where T is the sample, or reference, time, and ${\tau}$ is the gust duration, and a non-dimensional mean wind speed, $\bar{U}.T/L_u$, where $\bar{U}$ is a mean wind speed, and $L_u$ is the integral length scale of turbulence. The commonly-used Durst relationship, relating gusts of various durations, is shown to correspond to a particular value of turbulence intensity $I_u$, of 16.5%, and is therefore applicable to particular terrain and height situations, and hence should not be applied universally. The effective frontal areas associated with peak gusts of various durations are discussed; this indicates that a gust of 3 seconds has an equivalent frontal area equal to that of a tall building. Finally a generalized gust response factor format, accounting for fluctuating and resonant along-wind loading of structures, applicable to any code is presented.

• Journal of the Korea Computer Industry Society
• /
• v.5 no.5
• /
• pp.737-746
• /
• 2004

Two-channel multiwavelet system is investigated for image compression application in this paper. Generally, multiwavelets are known for their superb capability of compressing non-stationary signals like voice. However, multivavelet system have a critical problem in processing and compressing image data due to mesh-grid visual artifacts. In our two-channel multiwavelet system we have investigated incorporation of pre and post filtering to the multiwavelet transform and compression system for alleviating those ingerent visual artifacts due to multiwavelet effect. In addition, to quantify the image data compression performance of proposed multiwavelet system, computer simulations have been performed using various image data. For bit allocation and quantization, the Lagrange multiplier technique considering data rate vs. distortion rate along with a nonlinear companding method are applied equallly to all systems considered, here. The simulation results have yielded 1 ~ 2 dB compression enhancement over the scalar savelet systems. If the more advanced compression methods like SPIHT and run-length channel coding were adopted for the proposed multiwavelet system, a much higher compression gain could be obtained.

• Structural Engineering and Mechanics
• /
• v.38 no.1
• /
• pp.39-63
• /
• 2011

Referring to the formulation of physical stochastic optimal control of structures and the scheme of optimal polynomial control, a nonlinear stochastic optimal control strategy is developed for a class of structural systems with hysteretic behaviors in the present paper. This control strategy provides an amenable approach to the classical stochastic optimal control strategies, bypasses the dilemma involved in It$\hat{o}$-type stochastic differential equations and is applicable to the dynamical systems driven by practical non-stationary and non-white random excitations, such as earthquake ground motions, strong winds and sea waves. The newly developed generalized optimal control policy is integrated in the nonlinear stochastic optimal control scheme so as to logically distribute the controllers and design their parameters associated with control gains. For illustrative purposes, the stochastic optimal controls of two base-excited multi-degree-of-freedom structural systems with hysteretic behavior in Clough bilinear model and Bouc-Wen differential model, respectively, are investigated. Numerical results reveal that a linear control with the 1st-order controller suffices even for the hysteretic structural systems when a control criterion in exceedance probability performance function for designing the weighting matrices is employed. This is practically meaningful due to the nonlinear controllers which may be associated with dynamical instabilities being saved. It is also noted that using the generalized optimal control policy, the maximum control effectiveness with the few number of control devices can be achieved, allowing for a desirable structural performance. It is remarked, meanwhile, that the response process and energy-dissipation behavior of the hysteretic structures are controlled to a certain extent.

• Proceedings of the KSME Conference
• /
• 2001.11b
• /
• pp.324-329
• /
• 2001

This experimental study concerns the characteristics of a transitional flow in a concentric annulus with a diameter ratio of 0.52, whose outer cylinder is stationary and inner one rotating. The pressure losses and skin-friction coefficients have been measured for the fully developed flow of $0.1\sim0.4%$ aqueous solution of sodium carbomethyl cellulose (CMC), respectively at inner cylinder rotational speed of $0\sim600rpm$. The transitional flow has been examined by the measurement of pressure losses to reveal the relation of the Reynolds and Rossby numbers with the skin-friction coefficients. The present results show that the skin-friction coefficients have the significant relation with the Rossby numbers, only for laminar regime. The occurrence of transition has been checked by the gradient changes of pressure losses and skin-friction coefficients with respect to the Reynolds numbers. The increasing rate of skin-friction coefficients due to the rotation in uniform for laminar flow regime, whereas it is suddenly reduced for transitional flow regime and, then, is gradually declined for turbulent flow regime. Consequently, the critical(axial-flow) Reynolds number decrease as the rotational speed increases. Thus, the rotation of inner cylinder promotes the early occurrence of transition due to the onset of taylor vortices.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.15 no.5
• /
• pp.227-234
• /
• 2020

In this paper, we propose wav-U-Net to improve speech enhancement in heavy noisy environments, and it has implemented three principal techniques. First, as input data, we use 128 modified Mel-scale filter banks which can reduce computational burden instead of 512 frequency bins. Mel-scale aims to mimic the non-linear human ear perception of sound by being more discriminative at lower frequencies and less discriminative at higher frequencies. Therefore, Mel-scale is the suitable feature considering both performance and computing power because our proposed network focuses on speech signals. Second, we add a simple ResNet as pre-processing that helps our proposed network make estimated speech signals clear and suppress high-frequency noises. Finally, the proposed U-Net model shows significant performance regardless of the kinds of noise. Especially, despite using a single channel, we confirmed that it can well deal with non-stationary noises whose frequency properties are dynamically changed, and it is possible to estimate speech signals from noisy speech signals even in extremely noisy environments where noises are much lauder than speech (less than SNR 0dB). The performance on our proposed wav-U-Net was improved by about 200% on SDR and 460% on NSDR compared to the conventional Jansson's wav-U-Net. Also, it was confirmed that the processing time of out wav-U-Net with 128 modified Mel-scale filter banks was about 2.7 times faster than the common wav-U-Net with 512 frequency bins as input values.

• Transactions of the Korean Society for Noise and Vibration Engineering
• /
• v.20 no.11
• /
• pp.1078-1088
• /
• 2010

In order to analyze the quantitative characteristics of acoustic streaming, experimental setup of 3-D stereoscopic PIV(particle imaging velocimetry) was designed and quantitative ultrasonic flow fields in the gap between the ultrasonic vibrator and heat source were measured. Utilizing acoustic streaming induced by ultrasonic vibration, surface temperature drop of cooling object was also measured. The study on smart cooling method by acoustic streaming induced by ultrasonic vibration was performed due to the empirical relations of flow pattern, average flow velocity, different gaps, and enhancement on cooling rates in the gap. Average velocity fields and maximum acoustic streaming velocity in the open gap between the stationary cylindrical heat source and ultrasonic vibrator were experimentally measured at no vibration, resonance, and non-resonance. It was clearly observed that the enhancement of cooling rates existed owing to the acoustic air flow in the gap at resonance and non-resonance induced by ultrasonic vibration. The ultrasonic wave propagating into air in the gap creates steady-state secondary eddy called acoustic streaming which enhances heat transfer from the heat source to encompassing air. The intensity of the acoustic streaming induced by ultrasonic vibration experimentally depended upon the gap between the heat source and ultrasonic vibrator. The ultrasonic vibration at resonance caused the increase of the acoustic streaming velocity and convective heat transfer augmentation when the flow fields by 3D stereoscopic PIV and temperature drop of the heat source were measured experimentally. The acoustic streaming velocity of air enhancement on cooling rates in the gap is maximal when the gap agrees with the multiples of half wavelength of the ultrasonic wave, which is specifically 12 mm.

• Korean Journal of Food Science and Technology
• /
• v.17 no.3
• /
• pp.203-207
• /
• 1985

Relationship between cell surface hydrophobicity and pellicle formation was studied in a film strain isolated from stored apple wine and identified as Hansenula beijerinckii FY-5. In the media containing non-ionic surface-active agents the pellicle formation of strain FY-5 was efficiently repressed, whereas growth of the yeast was possible, and also cell surface hydrophobicity was greatly decreased by the addition of these agents. These results indicate that a pellicle formation factor, which keeps yeast cells floating on the medium surface, is necessary for the pellicle formation, and surely this factor is the hydrophobicity of the cell surface. The pellicle formation in the film strains was abundant with the increase of the cell surface hydrophobicity, whereas the non-film strains had less hydrophobicity as compared with the film strains. Ethanol, as a sole carbon source, efficiently increased hydrophobicity more than glucose, and the hydrophobicity was lowered with the rise of pH. In the experiments of time course, the hydrophobicity was increased in proportion to cell growth, and was maximum during the stationary phase.

• Journal of Ocean Engineering and Technology
• /
• v.23 no.4
• /
• pp.69-77
• /
• 2009

Structural analysis of a brake shoe for commercial vehicle was performed using finite element method. Since the strength of a brake shoe is affected by the magnitude and distribution shape of the contact pressure with the drum, the contact pressure between the shoe friction material and drum was calculated using a 2-Dimensional non-linear contact analysis in a state. And the brake was actuated by input air pressure and the drum of it was calculated both stationary and dynamic based on forced torque applied to the drum during the static state analysis. The results of the above analysis were then used as the load boundary conditions for a 3-Dimensional shoe model analysis to determine the maximum strain on the shoes. In the analysis model, the values of tensile test were used for the material properties of the brake shoes and drum, while the values of compression test were used for the friction material. We assumed it as linear variation, even though the properties of friction material were actually non-linear. The experiments were carried out under the same analysis conditions used for fatigue test and under the same brake system which equipped with a brake drum based on the actual axle state in a vehicle. The strains were measured at the same locations where the analysis was performed on the shoes. The obtained results of the experiment matched well with those from the analysis. Consequently, the model used in this study was able to determine the stress at the maximum air pressure at the braking system, thereby a modified shoe model in facilitating was satisfied with the required endurance strength in the vehicle.