• Title/Summary/Keyword: 성능평가모형

Search Result 853, Processing Time 0.03 seconds

자료기반 수문 예측모형의 성능평가 기법 개발

  • Hwang, Seok-Hwan;Kim, Jung-Hun;Jeong, Seong-Won
    • Proceedings of the Korea Water Resources Association Conference
    • /
    • 2010.05a
    • /
    • pp.1536-1540
    • /
    • 2010
  • 자료기반 수문예측 모형은 서로 자기상관이 다른 자료계열에 대해 예측결과만으로 모형의 성능에 대한 상대비교가 어렵다. 그러나 관측치와 예측치간의 평균 오차만을 기준으로 판단하는 기존의 모형 성능평가 기법은 대부분 이러한 자료기반 예측모형의 특성을 고려하지 못하고 있다. 따라서 본 논문에서는 자료기반 수문 예측모형의 성능을 보다 객관적으로 평가할 수 있는 새로운 모형 성능평가 기법인 상대 상관계수(Relative Correlation Coefficient; RCC) 제시하였다. RCC는 자기상관계수에 대한 관측치와 예측치간의 상관계수의 비로 산정되며, 자기상관정도에 따라 예측성능의 결과가 달라진다. 본 논문에서는 다양한 자기상관을 가지는 선형, 비선형 자료계열에 대해 자료기반 수문모형을 적용하여 기존 모형평가 기법의 한계를 제시하였다. 그리고 기존의 성능평가 기법과 RCC를 비교분석하여 자료기반 수문예측모형의 성능평가에 있어 RCC가 보다 객관적이고 일관성 있는 성능평가가 가능함을 보였다.

  • PDF

Seasonality Analysis of Soil Moisture using Cyclostationary Empirical Orthogonal Function (CSEOF 분석을 이용한 토양수분의 계절성 분석)

  • Cho, Eunsaem;Lee, Hyoungtaek;Lee, Myungseob;Lee, Youngju;Yoo, Chulsang
    • Proceedings of the Korea Water Resources Association Conference
    • /
    • 2016.05a
    • /
    • pp.282-282
    • /
    • 2016
  • 지표수문해석모형이란 전 지구를 대상으로 수문해석 및 예측이 가능한 분포형 수문모형이다. 본 연구에서는 CSEOF(Cyclostationary Empirical Orthogonal Functions) 분석 방법을 이용하여 지표수문해석 모형 중 하나인 VIC(Variable Infiltration Capacity)모형의 토양수분 모의 성능을 평가해보고자 한다. 이를 위하여 먼저 남한에 대한 VIC 모형으로 모의한 토양수분 예측 결과와 관측자료를 수집하였다. 모의 성능 평가 기간은 1976년부터 2006년까지이다. 이후 본 연구에서는 수집된 VIC 모형의 예측 결과와 관측 자료에 대한 CSEOF 분석을 수행하여 각 자료의 월별 주된 변동 특성을 추출하였다. VIC 모형의 예측 결과와 관측자료의 상관관계는 CSEOF 분석 결과에 대한 Pattern Correlation으로 정량화되었다. 이와 더불어 본 연구에서는 모형의 모의 성능 평가에 주로 사용되는 NRMSE(Nomalized Root Mean Square Error)를 산정하여 예측 결과의 오차를 평가하였다. Pattern Correlation과 NRMSE를 모두 고려하여 VIC 모형의 성능을 평가해본 결과, 건기에 해당하는 기간과 우기에 해당하는 기간의 모의 성능이 다르게 나타났다. 본 연구의 결과는 추후에 지표수문해석 모형의 예측 결과를 이용하는 기후변화 관련 연구에 활용될 수 있을 것으로 판단된다.

  • PDF

Pan evaporation modeling using deep learning theory (Deep learning 이론을 이용한 증발접시 증발량 모형화)

  • Seo, Youngmin;Kim, Sungwon
    • Proceedings of the Korea Water Resources Association Conference
    • /
    • 2017.05a
    • /
    • pp.392-395
    • /
    • 2017
  • 본 연구에서는 일 증발접시 증발량 산정을 위한 딥러닝 (deep learning) 모형의 적용성을 평가하였다. 본 연구에서 적용된 딥러닝 모형은 deep belief network (DBN) 기반 deep neural network (DNN) (DBN-DNN) 모형이다. 모형 적용성 평가를 위하여 부산 관측소에서 측정된 기상자료를 활용하였으며, 증발량과의 상관성이 높은 기상변수들 (일사량, 일조시간, 평균지상온도, 최대기온)의 조합을 고려하여 입력변수집합 (Set 1, Set 2, Set 3)별 모형을 구축하였다. DBN-DNN 모형의 성능은 통계학적 모형성능 평가지표 (coefficient of efficiency, CE; coefficient of determination, $r^2$; root mean square error, RMSE; mean absolute error, MAE)를 이용하여 평가되었으며, 기존의 두가지 형태의 ANN (artificial neural network), 즉 모형학습 시 SGD (stochastic gradient descent) 및 GD (gradient descent)를 각각 적용한 ANN-SGD 및 ANN-GD 모형과 비교하였다. 효과적인 모형학습을 위하여 각 모형의 초매개변수들은 GA (genetic algorithm)를 이용하여 최적화하였다. 그 결과, Set 1에 대하여 ANN-GD1 모형, Set 2에 대하여 DBN-DNN2 모형, Set 3에 대하여 DBN-DNN3 모형이 가장 우수한 모형 성능을 나타내는 것으로 분석되었다. 비록 비교 모형들 사이의 모형성능이 큰 차이를 보이지는 않았으나, 모든 입력집합에 대하여 DBN-DNN3, DBN-DNN2, ANN-SGD3 순으로 모형 효율성이 우수한 것으로 나타났다.

  • PDF

A study on performance evaluation method of advanced machine tools (자동화제조시스템용으로 개발되는 공작기계의 성능평가방법에 대한 연구)

  • 편영식
    • Proceedings of the Korean Operations and Management Science Society Conference
    • /
    • 1995.04a
    • /
    • pp.158-165
    • /
    • 1995
  • G7 프로젝트로 개발중인 첨단생산시스템(FMS로부터 CIM까지)의 핵심구성요소인 고성능.고정밀공작기계의 성능을 평가할 수 있는 요소를 개발하고 사례로서 이를 Machining Centers에 적용하여 각 요소별 세부항목과 기준을 개발하여 제안하였다. 또한 이 평가내용을 국제경쟁제품과 비교하거나 국제경쟁력 달성정도를 평가할 때 용이하게 사용할 수 있는 적절한 평가모형도 제안하였다.

  • PDF

A Study on Describing Uninterrupted Traffic Flows using Macroscopic Models (연속교통류 재현을 위한 거시적 모형의 비교 연구)

  • 임성만;김대호;김영찬
    • Journal of Korean Society of Transportation
    • /
    • v.20 no.3
    • /
    • pp.69-82
    • /
    • 2002
  • The objective of this study is to evaluate the performance of macroscopic traffic flow models with the analytical and field data. Five candidate models were selected as follows ; Lax Method Model, Upwind Scheme Model, Hilliges'Model, Papageorgiou's Model, and Cell-Transmission Model. In the analytical test scenario, the traffic condition was assumed that could cause the building and dissipation of queue, and each model was compared with analytical solutions and the numerical results. An analytical test indicated that both simple continuum and high order continuum models are able to reproduce queue building and dissipating behavior in a reasonable way A field test has shown that Upwind and Papageorgiou's model show similar performances. Considering the simplicity in model formulation and numerical computation, we firstly recommend Upwind scheme model , and secondly Papageorgiou's model that performed will to represent traffic flow in tests as candidate models for further development of simulation model for Naebu expressway in Seoul.

Comparisons of the corporate credit rating model power under various conditions (기준값 변화에 따른 기업신용평가모형 성능 비교)

  • Ha, Jeongcheol;Kim, Soojin
    • Journal of the Korean Data and Information Science Society
    • /
    • v.26 no.6
    • /
    • pp.1207-1216
    • /
    • 2015
  • This study aims to compare the model power in developing corporate credit rating models and to suggest a good way to build models based on the characteristic of data. Among many measurement methods, AR is used to measure the model power under various conditions. SAS/MACRO is in use for similar repetitions to reduce time to build models under several combination of conditions. A corporate credit rating model is composed of two sub-models; a credit scoring model and a default prediction model. We verify that the latter performs better than the former under various conditions. From the result of size comparisons, models of large size corporate are more powerful and more meaningful in financial viewpoint than those of small size corporate. As a corporate size gets smaller, the gap between sub-models becomes huge and the effect of outliers becomes serious.

Classification and Performance Evaluation Methods of an Algal Bloom Model (적조모형의 분류 및 성능평가 기법)

  • Cho, Hong-Yeon;Cho, Beom Jun
    • Journal of Korean Society of Coastal and Ocean Engineers
    • /
    • v.26 no.6
    • /
    • pp.405-412
    • /
    • 2014
  • A number of algal bloom models (red-tide models) have been developed and applied to simulate the redtide growth and decline patterns as the interest on the phytoplankton blooms has been continuously increased. The quantitative error analysis of the model is of great importance because the accurate prediction of the red-tide occurrence and transport pattern can be used to setup the effective mitigations and counter-measures on the coastal ecosystem, aquaculture and fisheries damages. The word "red-tide model" is widely used without any clear definitions and references. It makes the comparative evaluation of the ecological models difficult and confusable. It is highly required to do the performance test of the red-tide models based on the suitable classification and appropriate error analysis because model structures are different even though the same/similar words (e.g., red-tide, algal bloom, phytoplankton growth, ecological or ecosystem models) are used. Thus, the references on the model classification are suggested and the advantage and disadvantage of the models are also suggested. The processes and methods on the performance test (quantitative error analysis) are recommend to the practical use of the red-tide model in the coastal seas. It is suggested in each stage of the modeling procedures, such as verification, calibration, validation, and application steps. These suggested references and methods can be attributed to the effective/efficient marine policy decision and the coastal ecosystem management plan setup considering the red-tide and/or ecological models uncertainty.

Validity Analysis of Scale Model Experiment for Wetting Agent Performance Evaluation (침윤소화약제 성능평가를 위한 축소실험의 타당성 분석)

  • Kim, Nam-Kyun;Lim, Kyung-Bum;Rie, Dong-Ho
    • Fire Science and Engineering
    • /
    • v.28 no.2
    • /
    • pp.14-19
    • /
    • 2014
  • A current standard exist only on the surface tension in the current domestic wetting agent technology standards, so it is difficult to the performance evaluation of the wetting agent through the standard. So this study presents the optimized performance evaluation methods by scale model experimental equipment in order to present techniques for performance evaluation of wetting agents. The purpose of this study is to investigate validity of experimental results of the self-designed scale model experiment equipment by a comparative analysis of experimental results of the NFPA 18 experiment and the experiment using the self-designed scale model experiment equipment. As a result of a comparative analysis of experimental results of the NFPA 18 experiment that evaluate only the permeation performance on the contton and the experiment using the self-designed scale model experiment equipment that evaluate the permeation performance and fire extinguishing performance on wood flour, the discrimination of the permeation performance was confirmed in both the NFPA 18 experiment and the self-designed scale model experiment equipment. And a result of self-designed experiment equipment have clear discriminatory more than NFPA 18 by internal temperature measurement using the thermocouples.

A comparative study of conceptual model and machine learning model for rainfall-runoff simulation (강우-유출 모의를 위한 개념적 모형과 기계학습 모형의 성능 비교)

  • Lee, Seung Cheol;Kim, Daeha
    • Journal of Korea Water Resources Association
    • /
    • v.56 no.9
    • /
    • pp.563-574
    • /
    • 2023
  • Recently, climate change has affected functional responses of river basins to meteorological variables, emphasizing the importance of rainfall-runoff simulation research. Simultaneously, the growing interest in machine learning has led to its increased application in hydrological studies. However, it is not yet clear whether machine learning models are more advantageous than the conventional conceptual models. In this study, we compared the performance of the conventional GR6J model with the machine learning-based Random Forest model across 38 basins in Korea using both gauged and ungauged basin prediction methods. For gauged basin predictions, each model was calibrated or trained using observed daily runoff data, and their performance was evaluted over a separate validation period. Subsequently, ungauged basin simulations were evaluated using proximity-based parameter regionalization with Leave-One-Out Cross-Validation (LOOCV). In gauged basins, the Random Forest consistently outperformed the GR6J, exhibiting superiority across basins regardless of whether they had strong or weak rainfall-runoff correlations. This suggest that the inherent data-driven training structures of machine learning models, in contrast to the conceptual models, offer distinct advantages in data-rich scenarios. However, the advantages of the machine-learning algorithm were not replicated in ungauged basin predictions, resulting in a lower performance than that of the GR6J. In conclusion, this study suggests that while the Random Forest model showed enhanced performance in trained locations, the existing GR6J model may be a better choice for prediction in ungagued basins.

Assessing the Metric to Measuring Land-Use Change Suitability (토지 이용 변화 예측 모형의 정확도 검정을 위한 통계량 연구)

  • Kim, Oh Seok
    • Journal of the Economic Geographical Society of Korea
    • /
    • v.16 no.3
    • /
    • pp.458-471
    • /
    • 2013
  • This paper addresses the limitation of a map comparison metric entitled Figure of Merit through employing a simple land change model. The metric was originally designed to overcome limitations of other existing statistics, such as Kappa, when assessing predictive accuracy of land change models. A series of comparisons between null and predicted outcomes at multiple resolutions as well as a multi-resolution Figure of Merit analysis techniques of validation are compared for spatially segregated calibration and validation datasets. The Figure of Merit at the null resolution in this paper was 57%, although future research must be done to determine if this was simply a coincidence. A Figure of Merit greater than 50% would seem to represent a "Resolution of Merit" in that the Figure of Merit at that resolution becomes greater than the error. Thus, these two metrics should be used in tandem to assess predictive accuracy of a land change model.

  • PDF