Journal of Korea Water Resources Association (한국수자원학회논문집)

Korea Water Resources Association (한국수자원학회)
권호 표지

A Study on the Forecasting of Daily Streamflow using the Multilayer Neural Networks Model

다층신경망모형에 의한 일 유출량의 예측에 관한 연구

  • 김성원 (콜로라도 주립대학교 토목공학과)
  • Published : 2000.10.01
Download PDF
⟨ Previous Next ⟩

Abstract

In this study, Neural Networks models were used to forecast daily streamflow at Jindong station of the Nakdong River basin. Neural Networks models consist of CASE 1(5-5-1) and CASE 2(5-5-5-1). The criteria which separates two models is the number of hidden layers. Each model has Fletcher-Reeves Conjugate Gradient BackPropagation(FR-CGBP) and Scaled Conjugate Gradient BackPropagation(SCGBP) algorithms, which are better than original BackPropagation(BP) in convergence of global error and training tolerance. The data which are available for model training and validation were composed of wet, average, dry, wet+average, wet+dry, average+dry and wet+average+dry year respectively. During model training, the optimal connection weights and biases were determined using each data set and the daily streamflow was calculated at the same time. Except for wet+dry year, the results of training were good conditions by statistical analysis of forecast errors. And, model validation was carried out using the connection weights and biases which were calculated from model training. The results of validation were satisfactory like those of training. Daily streamflow forecasting using Neural Networks models were compared with those forecasted by Multiple Regression Analysis Mode(MRAM). Neural Networks models were displayed slightly better results than MRAM in this study. Thus, Neural Networks models have much advantage to provide a more sysmatic approach, reduce model parameters, and shorten the time spent in the model development.

본 연구에서는 낙동강 진동지점에서 일유출량을 예측하기 위하여 신경망모형이 제시되었다. 신경망모형의 구조는 CASE 1(5-5-1)과 CASE 2(5-5-5-1)로 구성하였으며, 은닉층의 수에 따라 두 가지의 모형으로 분류하였다. 각 신경망모형은 광역최소점과 훈련임계치에 수렴하는데 기존의 역전파훈련 알고리즘(BP) 보다 뛰어난 Fletcher-Reeves 공액구배 역전파훈련 알고리즘(FR-CGBP)과 축적된 공액구배 역전파훈련 알고리즘(SCGBP)을 이용하였다. 그리고 모형의 훈련과 검증을 위하여 이용된 자료는 풍수년, 평수년, 갈수년 풍수년+평수년, 풍수년+갈수년, 평수년+갈수년 및 풍수년+평수년+갈수년으로 구분하여 구성하였다. 모형의 훈련과정에서 각 자료를 이용하여 최적 연결강도와 편차가 결정되어 졌으며, 동시에 일유출량이 계산되어졌다. 예측오차의 통계분석을 통하여 풍수년+갈수년의 자료를 제외하고는 훈련결과가 양호한 것으로 나타났다. 모형의 검증에는 모형의 훈련을 통해 산정된 CASE 1 의 SCGBP 알고리즘의 연결강도와 편차를 이용하였으며, 검증의 결과는 훈련결과처럼 만족스러운 것으로 분석되었다. 또한 본 연구에서 선정한 신경망모형과 비교검토하기 위하여 다중회귀분석모형을 적용하여 일유출량을 예측하였으며, 그 결과 신경망모형이 다소 우수한 결과를 나타내는 것으로 분석되었다. 이와 같이 신경망모형은 조직적인 접근법, 매개변수의 감소 및 모델을 개발하는데 소모되는 시간을 줄일수 있는 장점이 있다.

Keywords

References

  1. 건설교통부， 한국수자원공사(1999) 기존댐 용수공급 능력조사(낙동강 . 금강수계)보고서
  2. 김주환(1993) '신경회로망을 이용한 하천유출량의 수문학적 예측에 관한 연구' 박사학위논문， 인하대학교
  3. 심순보， 김만식(1999a). '충주다목적댐 홍수유입량 예측을 위한 최적 신경망모형의 개발.' 대한토목학회논문집， 대한토목학회， 제 19권, 제II-1호，pp 67-78
  4. Demuth, H.. and Beale, M.(1998). Neural network toolbox : for use with MATLAB user's guide, The Math Works Inc
  5. Fletcher, R.(1975). Practical methods of optimization, John Wiley & Sons, New York
  6. Fletcher, R, and Reeves, C.M.(1964). 'Function minimization by conjugate gradient' Computer Journal, Vol. 7, pp. 149-154 https://doi.org/10.1093/comjnl/7.2.149.
  7. Hipel, KW., and McLeod, A.I.(1994). Time series modeling of water resources and environmental systems, Development in Water Science 45, Elsevier
  8. Isobe, I., Ohkado, T., Hanyuda, H., Oda, S., and Gotoh, Y.(1994). 'The development of a forecasting system of the water levels of rivers by neural networks.' J. of Japan Soc. of Hydro. and Water Resour., Vol. 7, No.2, pp. 90-97
  9. Kim, S., and Lee, S.(2000). 'Forecasting of flood stage using neural networks in the Nakdong river, South Korea.' Proc., Watershed Management & Operations Management 2000, ASCE/EWRI. Fort Collins, CO
  10. Liong, S.Y., and Chan, W.T.(l993). 'Runoff volume estimates with neural network.' Proc., 3rd Int. Conf. in Application of AI to Civ. and Struct. Engrg. Neural Networks and Combinatorial Optimization in Civ. and Struct. Engrg., B.H.V. Topping and A.I.Khan, eds., Civil-Comp. Press, U.K., pp. 67-70
  11. Minns, A.W., and Hall, M.J.(l996). 'Artificial neural network as rainfall runoff models.' J. of Hydr. Res., Vol. 34, No.4, pp. 537-548
  12. Moller, M.F.(1993). 'A scaled conjugate gradient algorithm for fast supervised learning' Neural Networks, Vol. 6, pp. 525-533 https://doi.org/10.1016/S0893-6080(05)80056-5
  13. NeuralWare, Inc., Neural Computing(1991). NeuralWorks Professional II/Plus and NeuralWorks Explorer, Pittsburgh, PA
  14. Nguyen, D.H. and Widrow, B,(1990). 'Neural network for self-learning control systems.' IEEE Control System, Magazine, pp. 18-23 https://doi.org/10.1109/37.55119
  15. NWS(National Weather Service)(1996). National Weather Service River Forecast System (NWSRFS), U.S. Department of Commerce, NOAA, National Weather Service, Office of Hydrology. Silver Spring, MD
  16. Raman, H., and Sunilkumar, N,(I995). 'Modeling water resources time series using artificial neural networks.' Hydro. Sci., Vol. 40, No.2, pp. 145-162
  17. Rosso, R, Peano, A, Becchi, I., and Bemporad, G.A.(1994). Advances in distributed hydrology, Water Resources Publications, Littleton, CO
  18. Salas, J.D., Smith, R.A, Tabios III, G.Q., and Heo, J.H. (1999). Statistical computer techniques in water resources and environmental engineering, Unpublished text book in CE622, Colorado State University, Fort Collins, CO
  19. Smith, J., and Eli, R.N.(1995), 'Neural network models of rainfall-runoff process.' J. of Water Resour. Plng. and Mgmt., ASCE, Vol. 121, No.6, pp. 499-508 https://doi.org/10.1061/(ASCE)0733-9496(1995)121:6(499)
  20. Thirumalaiah, K., and Deo, M.C.(1998). 'River stage forecasting using artificial neural networks.' J. of Hydro. Eng., ASCE, Vol. 3, No.1, pp. 26-32 https://doi.org/10.1061/(ASCE)1084-0699(1998)3:1(26)
  21. Tokar, A.S., and Johnson, P.A(1999). 'Rainfall-runoff modeling using artificial neural networks.' J. of Hydro. Eng., ASCE, Vol. 4, No.3, pp. 232-239 https://doi.org/10.1061/(ASCE)1084-0699(1999)4:3(232)
  22. Zealand, C.M., Burn, D.H., and Simonovic, S.P.(1999). 'Short term streamflow forecasting using artificial neural networks.' J. of Hydro., Vol. 214, pp. 32-48 https://doi.org/10.1016/S0022-1694(98)00242-X