• Title/Summary/Keyword: Loss

Search Result 34,569, Processing Time 0.068 seconds

ACL-GAN: Image-to-Image translation GAN with enhanced learning and hyper-parameter searching speed using new loss function (ACL-GAN: 새로운 loss 를 사용하여 하이퍼 파라메터 탐색속도와 학습속도를 향상시킨 영상변환 GAN)

  • Cho, JeongIk;Yoon, Kyoungro
    • Proceedings of the Korean Society of Broadcast Engineers Conference
    • /
    • 2019.11a
    • /
    • pp.41-43
    • /
    • 2019
  • Image-to-image 변환에서 인상적인 성능을 보이는 StarGAN 은 모델의 성능에 중요한 영향을 끼치는 adversarial weight, classification weight, reconstruction weight 라는 세가지 하이퍼파라미터의 결정을 전제로 하고 있다. 본 연구에서는 이 중 conditional GAN loss 인 adversarial loss 와 classification loss 를 대치할 수 있는 attribute loss를 제안함으로써, adversarial weight와 classification weight 를 최적화하는 데 걸리는 시간을 attribute weight 의 최적화에 걸리는 시간으로 대체하여 하이퍼파라미터 탐색에 걸리는 시간을 획기적으로 줄일 수 있게 하였다. 제안하는 attribute loss 는 각 특징당 GAN 을 만들 때 각 GAN 의 loss 의 합으로, 이 GAN 들은 hidden layer 를 공유하기 때문에 연산량의 증가를 거의 가져오지 않는다. 또한 reconstruction loss 를 단순화시켜 연산량을 줄인 simplified content loss 를 제안한다. StarGAN 의 reconstruction loss 는 generator 를 2 번 통과하지만 simplified content loss 는 1 번만 통과하기 때문에 연산량이 줄어든다. 또한 이미지 Framing 을 통해 배경의 왜곡을 방지하고, 양방향 성장을 통해 학습 속도를 향상시킨 아키텍쳐를 제안한다.

  • PDF

The Effect of the Processing Conditions on the Magnetics Properties of Mn-Zn ferrite (제조공정에 따른 Mn-Zn 페라이트의 전자기적 특성변화)

  • 김종령;이해연;김현식;오영우;민복기
    • Proceedings of the Korean Institute of Electrical and Electronic Material Engineers Conference
    • /
    • 2001.07a
    • /
    • pp.905-908
    • /
    • 2001
  • The microstructure and the magnetic properties of Mn-Zn ferrite, which were power loss and saturation magnetic flux density, were investigated as the function of the process before firing. The highest initial permeability and the lowest power loss were attained to the specimen with CaO 400 ppm as a resulted from the highest solubility to SiO$_2$and the creation of liquid phase which improved sintering. The biggest grain size, the highest saturation magnetic flux density and the lowest power loss, which was resulted from that the eddy current loss increased as grain size increased but the hysteresis loss much more decreased and the hysteresis loss strongly influenced on the total power loss rather than the eddy current loss, were obtained to the Mn-Zn ferrite added 2wt% PVA. The power loss was lowest and the saturation magnetic flux density was highest in case of 1 ton/$\textrm{cm}^2$ and the grain size was not influenced.

  • PDF

Estimation of Water Loss in Irrigation Canals through Field Measurement (현장 측정을 통한 관개용수로의 손실량 추정)

  • Lee, Yong-Jig;Kim, Phil-Shik;Kim, Sun-Joo;Keun, Jee-Yong;Joo, Uk-Jong
    • Journal of The Korean Society of Agricultural Engineers
    • /
    • v.50 no.1
    • /
    • pp.13-21
    • /
    • 2008
  • Water losses in irrigation canals are mainly estimated as the sum of conveyance and delivery water loss. The losses occur via the evaporation, infiltration, gate operation and water distribution processing. Recently, the study regarding these water losses are not satisfactory enough, also delivery water loss has not been mainly considered on field design. The objective of this study is to investigate and analyze the volume of water loss in irrigation canals considering condition of actual farm land. A field measurement was performed at four research sites, which are managed by Korea Rural Community & Agriculture Corporation, to evaluate conveyance and delivery water loss for 2 years. The measurement was performed by canal type, size and designed flow using the inflow-outflow method at a major points such as start and end of each canal, derivation point of canal and inlet of paddy fields. Results of this study showed that water loss ratio in lateral canals was bigger than that of main canal unlike current design standard and the loss decrease as flow increase. The total of water loss ratio including conveyance and delivery water loss in several irrigation canals ranged between 33.25 and 45.0%.

Analysis of Flow Rate Inducing Voltage Loss in a 100 cm2 Class Molten Carbonate Fuel Cell

  • Lee, Choong-Gon
    • Journal of Electrochemical Science and Technology
    • /
    • v.2 no.1
    • /
    • pp.20-25
    • /
    • 2011
  • This work focuses on the behavior of the overpotential increase due to a utilization rise in a molten carbonate fuel cell. The behavior is generally explained by Nernst loss, which is a kind of voltage loss due to the thermodynamic potential gradients in a polarization state due to the concentration distribution of reactant species through the gas flow direction. The evaluation of Nernst loss is carried out with a traditional experimental method of constant gas utilization (CU). On the other hand, overpotential due to the gas-phase mass-transport resistance at the anode and cathode shows dependence on the utilization, which can be measured using the inert gas step addition (ISA) method. Since the Nernst loss is assumed to be due to the thermodynamic reasons, the voltage loss can be calculated by the Nernst equation, referred to as a simple calculation (SC) in this work. The three values of voltage loss due to CU, ISA, and SC are compared, showing that these values rise with increases in the utilization within acceptable deviations. When we consider that the anode and cathode reactions are significantly affected by the gas-phase mass transfer, the behavior strongly implies that the voltage loss is attributable not to thermodynamic reasons, namely Nernst loss, but to the kinetic reason of mass-transfer resistance in the gas phase.

Magnetization Loss Characteristics of a Stacked Bi-2223 Conductor (적층 Bi-2223도체의 자화손실 특성)

  • Ryu, Gyeong-U;Han, Hyeong-Ju;Choe, Byeong-Ju;Na, Wan-Su;Ju, Jin-Ho
    • The Transactions of the Korean Institute of Electrical Engineers B
    • /
    • v.51 no.10
    • /
    • pp.554-559
    • /
    • 2002
  • The at loss is an important issue in the design of superconducting cables and transformers. In these devices the Bi-2222 tapes are usually placed face-to-face. In such arrangements ac loss is influenced by adjacent tapes. The effect is investigated by measuring the magnetization loss in the stacked conductor, which consists of various numbers of Bi-2223 tapes. For the single tape the magnetization loss in perpendicular field is larger than that in parallel field by about a factor 10. This agrees well with the prediction for hysteresis loss in slab and strip models. For the stacked conductor in perpendicular field the magnetization loss at low fields is greatly decreased, compared to the loss of the single tape. However the loss at high fields is nearly unaffected. This behavior is well described by the slab model.

A Study on the Quasi-3-Dimensional Compressible Flow Calculation by Introduction of Viscous Loss Model in Axial-Flow Compressor (점성 손실모델 도입에 의한 축류 압축기 준 3차원 압축성 유동해석)

  • 조강래;이진호;김주환
    • Transactions of the Korean Society of Mechanical Engineers
    • /
    • v.13 no.5
    • /
    • pp.1044-1051
    • /
    • 1989
  • A numerical calculation is carried out for the analysis of 3-dimensional compressible flow field in axial-flow rotating blades by using finite element method. The calculation of flow in impellers plays a dominant role in the theoretical research and design of turbomachines. Three-dimensional flow fields can be obtained by the quasi-three-dimensional iterative calculation of the flows both on blade-to-blade stream surfaces and hub-to-shroud stream surfaces with the introduction of viscous loss model in order to consider a loss due to viscosity of fluid. In devising the loss model, four primary sources of losses were identified: (1) blade profile loss (2) end wall loss (3) secondary flow loss (4) tip-leakage loss. For the consideration of an axially parabolic distribution of loss, the results of present calcullation are well agreed with the results by experiment, thus the introduction of loss model is proved to be valid.

Chaff-outlet Grain Loss of Head-feed Combine -Changes in chaff-outlet gram loss on the variety, location of chaff-outlet loss control plate and traveling speed- (자탈형(自脱型) 콤바인의 배진손실(排塵損失)에 관(關)한 연구(硏究)(II) -품종(品種), 배진조절판(排塵調節板)의 위치, 작업속도(作業速度)가 배진손실량(排塵損失量)에 미치는 영향-)

  • Chung, C.J.;Choe, J.S.;Choi, Y.S.;Chung, S.I.
    • Journal of Biosystems Engineering
    • /
    • v.15 no.4
    • /
    • pp.310-318
    • /
    • 1990
  • The amount of grain loss incurred during harvesting operation through the chaff-outlet of combine may not be negligible. To minimize this grain loss and optimize condition of combining, it may be necessary that the amout of chaff-outlet loss dependent on varying crop condition is to be estimated as exactly as possible. This study was thus intended to develop the monitor that could indicate the amount of grain lost through the chaff-outlet of combine during haravesting operation, and to find out driving and operating method of combine that could reduce chaff-outlet loss. In the study(1), the frequency distributions of each sound generated by the impact of kernels and chaff, the sound generated by the impact of the mixture of kernels and chaff, and vibration or noise created by the combine engine and other moving parts were investigated experimentally. Based on the results of frequency analyses, the loss monitor was developed which could measure the impact sound of paddy kernels that could be distinguished from those of other threshing products through chaff-outlet, and from vibration or noise created by the combine engine and other moving parts. Also in this study, detecting capability of monitor was tested by comparing the amount of grain lost through chaff-outlet with the amount of grain detected by the loss monitor, and changes in chaff-outlet grain loss on the increase of traveling speed, location of chaff-outlet loss control plate and variety of paddy rice were measured using the loss monitor. The monitor developed in this study efficiently measured the amount of grain lost through the chaff-outlet of combine. It was found that the chaff-outlet grain loss ratio was affected greatly by the variety of paddy rice, the location of chaff-outlet loss control plate and traveling speed of combine.

  • PDF

Numerical Analysis on Performance Changes of the Tubular SOFCs according to Current Collecting Method (전류집전 방법에 따른 원통형 고체산화물 연료전지의 성능 변화 수치해석)

  • Yu, Geon;Park, Seok-Joo;Lee, Jong-Won;Lee, Seung-Bok;Lim, Tak-Hyoung;Song, Rak-Hyun;Shin, Dong-Ryul;Kim, Ho-Young
    • Transactions of the Korean hydrogen and new energy society
    • /
    • v.22 no.2
    • /
    • pp.129-138
    • /
    • 2011
  • Performance changes of an anode-supported tubular SOFC including current collectors are analyzed at different current collecting methods using numerical simulation. From the two dimensional numerical model of the solid oxide fuel cell with nickel felts as anodic current collectors and silver wires as cathodic ones, the performance curves and the distributions of temperature, concentration, current density are obtained. Also, the voltage loss of the cell is divided into three parts: activation loss, concentration loss and ohmic loss. The results show that the performance change of the cell is dominantly influenced by the ohmic loss. Although the temperature and concentration distributions are different, the total activation loss and concentration loss are nearly same. And the ohmic loss is divided into each parts of the cell components. The ohmic loss of the anodic current collectorreaches about 60~80% of the cell's total ohmic loss. Therefore, the reduction of the ohmic loss of the anodic current collector is very important for stack power enhancement. It is also recommended that the load should be connected to the both ends of the anodic current collector.

An Experimental Study for Drainage Capacity Increment at Surcharged Manholes with a 90° Bend (과부하 90° 접합맨홀의 배수능력 증대에 관한 실험 연구)

  • Kim, Jung Soo;Song, Ju Il;Yoon, Sei Eui
    • Journal of Korean Society of Water and Wastewater
    • /
    • v.23 no.4
    • /
    • pp.447-458
    • /
    • 2009
  • Energy loss at manholes, often exceeding friction loss of pipes under surcharged flow, is considered as one of the major causes of inundation in urban areas. Therefore, it is necessary to analyze head loss at manholes, especially in case of surcharged flow. Hydraulic experiments were conducted with three cases. Case A is to test whether the shapes of the manholes influence head loss coefficients. Case B and C were proposed to further reduce head losses by improving the manhole hydraulic efficiency. In case B, the joining part of the pipe at both shapes of manholes is shifted from central part to side part. The test in case C is to check the average head loss coefficient by installing the side benching in square manhole, based on shifted joining part model. The average head loss coefficient for circular and square manhole on case A was 1.6. This did not show much difference of the head loss coefficients in spite of the discharge variation in this case. However, case B and C show large difference between head loss coefficients due to the strong oscillation of water surface and the horizontal swirl motion. The circular and square manholes in case B reduced the head loss by 30% and 6% than ones in case A, respectively. The average head loss coefficient for circular manhole in case B was 1.1. Case C reduced average loss coefficients of the square manhole in case A from 1.6 to 1.1. Accordingly, the circular manhole in case B and the square manhole in case C showed the effective way to reduce the head loss. These head loss coefficients could be available to apply to the urban sewer system with surcharged flow.

Optimum Welding Position between Shell and Cylinder based on SEA (SEA 를 이용한 쉘과 실린더의 최적 용접 조건)

  • Ahn, Byoung-Ha;Lee, Jang-Woo;Jeon, Simon;Yang, Bo-Suk
    • Proceedings of the Korean Society for Noise and Vibration Engineering Conference
    • /
    • 2012.10a
    • /
    • pp.258-264
    • /
    • 2012
  • The overall aim of this paper is to determine coupling loss factor of welding point between shell and cylinder using loss factor and structural loss factor. For this purpose, two kinds of loss factor were adopted. One is loss factor of each sub structure, another is structural loss factor based on the complex welded or assembled structure. Using these two parameters, it is possible to derive the coupling loss factor which represents characteristic condition of SEA theory. Coupling loss factor of conjunction in complex structure was expressed as power balance equation. The derived equation for a coupling loss factor has been simplified on the assumption of one way (uni-directional) power flow between multi-sub structures. Using these conditions, it is possible to find the equation of coupling loss factor expressed as above two loss factors. To check the effectiveness of above equation, this paper used two-stage application. The first approach was application between simple cylinder and shell. The next was adopted rotary compressor. Rotary compressor has three main conjunctions between shell and internal vibration part. This equation was applied to find out the optimum welding point with respect to reduce the noise propagation. It shows the effective tool to evaluate the coupling loss factor in complex structure.

  • PDF