• 제목/요약/키워드: minor loss

검색결과 187건 처리시간 0.022초

낙엽송(落葉松)의 종자결실(種子結實)에 영향(影響)을 미치는 요인(要因) (Factors Affecting Seed Yield in Larix)

  • 신동일;데이비드 F. 카노스키
    • 한국산림과학회지
    • /
    • 제84권2호
    • /
    • pp.207-217
    • /
    • 1995
  • 4가지 낙엽송(구주낙엽송, 일본낙엽송, 시베리아낙엽송, tamarack)의 종자결실에 영향을 미치는 요인과 각각의 요인들에 의한 종자손실율을 조사하였다. 화분의 질, 수분의 실패, female gametrophyte의 퇴화 등은 종자결실에 큰 영향을 미치지 않았으나, 수정실패, 배의 퇴화 등은 조사된 4개의 수조에서 모두 주요한 종자손실요인으로 조사되었다. 조사된 요인들 중 종자수확에 가장 큰 영향을 미치는 것은 수분후 strobilus자체가 퇴화되는 것이었다. 조사된 4가지 교배조합중 구주낙엽송 또는 일본낙엽송을 모수로 하는 것이 종자생산측면에서는 가장 바람직한 것으로 나타났으며, tamarack을 모수로 하는 교배조합은 건전한 종자생산을 하지 못할 것으로 판단된다.

  • PDF

A Robust Energy Consumption Forecasting Model using ResNet-LSTM with Huber Loss

  • Albelwi, Saleh
    • International Journal of Computer Science & Network Security
    • /
    • 제22권7호
    • /
    • pp.301-307
    • /
    • 2022
  • Energy consumption has grown alongside dramatic population increases. Statistics show that buildings in particular utilize a significant amount of energy, worldwide. Because of this, building energy prediction is crucial to best optimize utilities' energy plans and also create a predictive model for consumers. To improve energy prediction performance, this paper proposes a ResNet-LSTM model that combines residual networks (ResNets) and long short-term memory (LSTM) for energy consumption prediction. ResNets are utilized to extract complex and rich features, while LSTM has the ability to learn temporal correlation; the dense layer is used as a regression to forecast energy consumption. To make our model more robust, we employed Huber loss during the optimization process. Huber loss obtains high efficiency by handling minor errors quadratically. It also takes the absolute error for large errors to increase robustness. This makes our model less sensitive to outlier data. Our proposed system was trained on historical data to forecast energy consumption for different time series. To evaluate our proposed model, we compared our model's performance with several popular machine learning and deep learning methods such as linear regression, neural networks, decision tree, and convolutional neural networks, etc. The results show that our proposed model predicted energy consumption most accurately.

Cascaded Residual Densely Connected Network for Image Super-Resolution

  • Zou, Changjun;Ye, Lintao
    • KSII Transactions on Internet and Information Systems (TIIS)
    • /
    • 제16권9호
    • /
    • pp.2882-2903
    • /
    • 2022
  • Image super-resolution (SR) processing is of great value in the fields of digital image processing, intelligent security, film and television production and so on. This paper proposed a densely connected deep learning network based on cascade architecture, which can be used to solve the problem of super-resolution in the field of image quality enhancement. We proposed a more efficient residual scaling dense block (RSDB) and the multi-channel cascade architecture to realize more efficient feature reuse. Also we proposed a hybrid loss function based on L1 error and L error to achieve better L error performance. The experimental results show that the overall performance of the network is effectively improved on cascade architecture and residual scaling. Compared with the residual dense net (RDN), the PSNR / SSIM of the new method is improved by 2.24% / 1.44% respectively, and the L performance is improved by 3.64%. It shows that the cascade connection and residual scaling method can effectively realize feature reuse, improving the residual convergence speed and learning efficiency of our network. The L performance is improved by 11.09% with only a minimal loses of 1.14% / 0.60% on PSNR / SSIM performance after adopting the new loss function. That is to say, the L performance can be improved greatly on the new loss function with a minor loss of PSNR / SSIM performance, which is of great value in L error sensitive tasks.

지하 철근콘크리트 박스구조물의 균열제어 시공기술개발 (Development of Construction Methods for the Crack Control of underground RC Box Structures)

  • 이순환;김영진;김성운;방재원;최용성
    • 한국콘크리트학회:학술대회논문집
    • /
    • 한국콘크리트학회 1999년도 학회창립 10주년 기념 1999년도 가을 학술발표회 논문집
    • /
    • pp.809-812
    • /
    • 1999
  • This research analyzed the factors for crack generation and proposed the recommended construction methods for the efficient crack control of underground RC box structures under the roadway. The selected main factors were: details of contraction joints, ratio of crack control rebars in longitudinla direction, and placement of flyash concrete. These factors were tested on the actual structures and the significance of each factor was analyzed, The results show that the flyash concrete placement and the inducting minor cracks in a certain direction by adopting contraction joints are practical and efficient methods to control cracks. The significance of crack generating factors increases as the sectional loss of contraction joint spacing increase. It was recommeded that the sectional loss should be higher than 20 percent to maximize the crack generating effects. It was not possible to verify the effect of crack control rebar spacing, but it was estimated that the ratio of crack control rebar should be increased to minimize cracks.

  • PDF

항공기용 엔진제어기의 진공 브레이징 냉각유로 설계 및 압력손실 평가 (Design and Pressure Loss Evaluation of Vacuum Brazed Cooling Passage for Full Authority Digital Engine Control)

  • 한명재;설진운;정승호;차민경;장호연;김중회
    • 한국추진공학회지
    • /
    • 제26권2호
    • /
    • pp.72-78
    • /
    • 2022
  • 항공기용 엔진제어기는 주어진 환경에서 엔진의 최대 효율로 안전하게 운영될 수 있도록 엔진의 추력을 조절하고, 다른 보기 시스템들의 상태 감시를 수행하여 엔진의 모든 권한을 전자식으로 통합 제어하는 장치이다. 엔진제어기는 매우 높은 온도 환경에서도 정상 작동해야 한다. 따라서 엔진제어기는 내부 발열과 외부 유입 열을 고려한 최적의 방열설계가 필수적이다. 본 논문에서는 엔진제어기의 진공 브레이징 냉각유로를 설계하였다. 냉각유로의 전체 압력손실을 계산하기 위해 기본 형상에 대한 주손실과 입출구의 급격 확대/축소부, 유로 선회를 위한 밴드부 등의 비선형 형상에 대한 부차적손실을 계산하였다. 압력손실 이론식과 전산유체역학(Computational Fluid Dynamics, CFD) 해석을 활용한 합성추정법을 소개하여 각 비선형 형상에 대한 손실계수 계산하였다.

경증 두부 외상을 가진 환자의 경추 손상을 예측할 수 있는 관련 인자 (Predictive Factors for Cervical Spine Injury in Patients with Minor Head Injury)

  • 박철우;성애진;이준호;황성연
    • Journal of Trauma and Injury
    • /
    • 제22권2호
    • /
    • pp.154-160
    • /
    • 2009
  • Purpose: This study aimed to determine new criteria for detecting independent factors with high sensitivity in cases of cervical spine injury. We compared the sensitivity, the specificity, and the false negative predictive value (NPV) of plain radiographs with those of computed tomography for cervical spine injury in patients with minor head injury. Methods: We retrospectively reviewed the cases of 357 patients who underwent both cervical plain radiographs and computer tomography from January 2006, to September 2008. Patients were divided into two groups: the cervical spine injury group and the no cervical spine injury group. New criteria were organized based on variables that had significant differences in the logistic regression test. Results: Among the 357 patients, 78 patients had cervical spine injuries. The average age was $43.9{\pm}15.2$ yrs old, and the male-to-female ratio was 1.90. The most common mechanism of injury was motor vehicle accidents. There was a significant difference in loss of consciousness, Glasgow Coma Scale (GCS)=14, neurologic deficit, posterior neck tenderness, and abnormality of the cervical plain radiographs between the two groups on the logistic regression test. New criteria included the above five variables. If a patient has at least variable, the area under the ROC curve of the new criteria was 0.850, and the sensitivity and the false NPV were 87.2% and 5.2%, respectively. Conclusion: New criteria included loss of consciousness, GCS=14, neurologic deficit, posterior neck tenderness, and abnormality of the cervical plain radiographs. If the patient had at least 1 variable, he or she could have a of cervical spine injury with a sensitivity of 87.2% and a false NPV of 5.2%.

1.2MW급 산업용 가스터빈 원심압축기 개발(1)- 공력설계해석 - (Development of Centrifugal Compressors in an 1.2MW Industrial Gas Turbine(I)-Aerodynamic Design and Analysis-)

  • 조규식;이헌석;손정락
    • 대한기계학회논문집B
    • /
    • 제20권8호
    • /
    • pp.2707-2720
    • /
    • 1996
  • The aerodynamic design of the two-stages of centrifugal compressors in an 1.2MW industrial gas turbine is completed with the application of numerical analyses. The final shape of an intake, the axial guide vanes and a return channel is determined using several interactions between design and two-dimensional turbulent flow analysis, focused on the minimum loss of internal flows. The one-dimensional turbulent flow analysis, focused on the minimum loss of internal flows. The one-dimensional design and prediction of aerodynamic performances for the compressors are performed by two different methods; one is a method with conventional loss models, and the other a method with the two-zone model. The combination methods of the Betzier curves generate three-dimensional geometric shapes of impeller blades which are to be checked with a careful change of aerodynamic blade loadings. The impeller design is finally completed by the applications of three-dimensional compressible turbulent flow solvers, and the effect of minor change of design of the second-stage channel diffuser is also studied. All the aerodynamic design results are soon to the verified by component performance tests of prototype centrifugal compressors.

고온초전도선재의 교류손실 해석 및 실험 (AC loss analysis and experimental evaluation of a high temperature superconductor)

  • 류경우
    • 대한전기학회:학술대회논문집
    • /
    • 대한전기학회 2001년도 추계학술대회 논문집 전력기술부문
    • /
    • pp.96-100
    • /
    • 2001
  • Bi-2223 tapes have been developed for low-field power applications at liquid nitrogen temperature. When the Bi-2223 tapes are used in an application such as a power transmission cable or a power transformer, they are supplied with an AC transport current and exposed to an external magnetic field generated by neighboring tape's AC currents simultaneously. AC loss taking into account such real applications is a crucial issue for power applications of the Bi-2223 tapes to be feasible. In this paper, the transport losses for different AC current levels and arrangements of the neighboring tapes have been measured in a 1.5 m long Bi-2223 tape. The significant increase of the transport losses due to neighboring tape's AC currents is observed. An increase of the transport losses caused by a decrease of the Bi-2223 tape's critical current is a minor effect. The measured transport losses could not be explained by a dynamic resistance loss based on DC voltage-current characteristics in combination with the neighboring tape's AC currents. The transport losses do not depend on the frequency of the neighboring tape's AC currents but its arrangements in the range of small current especially.

  • PDF

Column-loss response of RC beam-column sub-assemblages with different bar-cutoff patterns

  • Tsai, Meng-Hao;Lua, Jun-Kai;Huang, Bo-Hong
    • Structural Engineering and Mechanics
    • /
    • 제49권6호
    • /
    • pp.775-792
    • /
    • 2014
  • Static loading tests were carried out in this study to investigate the effect of bar cutoff on the resistance of RC beam-column sub-assemblages under column loss. Two specimens were designed with continuous main reinforcement. Four others were designed with different types of bar cutoff in the mid-span and/or the beam-end regions. Compressive arch and tensile catenary responses of the specimens under gravitational loading were compared. Test results indicated that those specimens with approximately equal moment strength at the beam ends had similar peak loading resistance in the compressive arch phase but varied resistance degradation in the transition phase because of bar cutoff. The compressive bars terminated at one-third span could help to mitigate the degradation although they had minor contribution to the catenary action. Among those cutoff patterns, the K-type cutoff presented the best strength enhancement. It revealed that it is better to extend the steel bars beyond the mid-span before cutoff for the two-span beams bridging over a column vulnerable to sudden failure. For general cutoff patterns dominated by gravitational and seismic designs, they may be appropriately modified to minimize the influence of bar cutoff on the progressive collapse resistance.

Karhunen-Loeve 변환 기반의 부분공간 인식기와 결합된 다중 노벨티 인식기를 이용한 필기체 숫자 인식 (Handwritten Numeral Recognition Using Karhunen-Loeve Transform Based Subspace Classifier and Combined Multiple Novelty Classifiers)

  • 임길택;진성일
    • 전자공학회논문지C
    • /
    • 제35C권6호
    • /
    • pp.88-98
    • /
    • 1998
  • 부분공간 인식기는 Karhunen-Loeve (KL) 변환을 기반으로 하는 대표적인 패턴인식 방법이다. 이 부분 공간 인식기는 고차원의 패턴을 저차원의 부분공간에 나타내어 인식을 한다. 그러나 차원 감축으로 인한 정보의 손실로 principal components가 유사하게 나타나는 패턴간에는 분별이 어려워지는 단점이 있다. 본 논문에서는 이러한 부분공간 인식기의 단점을 해결하기 위해 일반적으로 무시되는 minor components로 표현되는 패턴의 노벨티 성분을 이용하는 결합된 다중 노벨티 신경망 인식기를 제안하고 부분공간 인식기와 결합을 통해 인식률을 제고하는 방법을 제시한다. 필기체 숫자 데이터베이스에 대해서 제안한 인식기를 구성하고 특성을 분석한다. 제안한 방법은 다른 인식기들에 비해서 신경망에 사용된 가중치의 수는 증가하지만 가장 우수한 인식 성능을 나타내었다.

  • PDF