The Runoff Forecasting Model Based on Wavelet Adaptive Neural Fuzzy Inference System

Abstract:

Considering various the wavelet decomposition reconstruction technology and training cycle of adaptive neural fuzzy inference system, this article propose four runoff forecast model of wavelet analysis and adaptive neural fuzzy inference system integration, such as the long cycle based on Mallat algorithm in runoff prediction, the long cycle based on wavelet packet algorithm in runoff prediction, the short cycle based on Mallat algorithm in runoff prediction, the short cycle based on wavelet packet algorithm in runoff prediction, and illuminate the model of the principles, structures and procedures. This model is used in Tangnaihe station monthly runoff forecast which lies in the Huanghe source area. Simulation results are evaluated by the cycle decomposition coefficients and Nash-Sutcliffe coefficient; it shows that the long cycle based on Mallat algorithm is best, the short cycle based on wavelet packet algorithm is worst. The author analyzes the reason and makes some proposal.

• Periodical: Applied Mechanics and Materials (Volumes 182-183)
• Pages: 1032-1040
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.182-183.1032
• Online since: June 2012
• Authors: Jun Fang Gu, Xiao Li Wang, Fu Hui Du
• Keywords: Adaptive Neural Fuzzy, Inference System, Runoff Forecast Model, Wavelet Analysis

References

[1] Li Xian-bin, Ding Jing, Li Hou-qiang, Journal of Hydraulic Engineering, vol. 2(1999), pp.1-4.

[2] Wang Wen-sheng, Xiong Hua-kang, Ding Jing. Advances In Water Science, vol. 15 (2004), pp.382-386.

[3] Guo Qi-yi, Lu Xiang-yang, Li Wei-Gang. Journal of Tongji University, 2005, Vol. 33 (2005), pp.130-133.

[4] Nash J E, Sutcliffe J V. Journal of Hydrology, Vol. 10(1997), pp.282-290.