Author(s)

Aimin Ning, Cunji Zhang and Yue Zhang

Corresponding Author

Cunji Zhang

ABSTRACT

To realize the accurate prediction for the flood flow, the prediction approach of the flood flow based on neuro-fuzzy networks is proposed. In recent years, the artificial intelligence has developed rapidly and applied in different fields widely. At first, the architecture of neuro-fuzzy networks is put forward in this paper, the parameters are adjusted by back-propagation iterative algorithm and the first order gradient optimization algorithm. For example, based on the measurement data at Nanning hydrological station in July 2001, the water level and the fluctuation rate are input into neuro-fuzzy networks, the flood flow is the output, the training and checking are performed by neuro-fuzzy networks, the results show that the neuro-fuzzy networks has high accuracy and low error in the prediction of flood flow.

KEYWORDS

Neuro-fuzzy networks, back-propagation iterative algorithm, the first order gradient optimization algorithm, flood flow, membership function