A Fuzzy-pattern Approach to Flood Classifying and Predicting

摘要

Scientific flood management is essential to predict and master some flood characteristics in real-time flood control dispatching. Previously, statistic analysis and neural networks were used in quantitative flood forecast, which is insufficient for real-time flood control operation. The objective of this study is to apply systemically fuzzy theory to flood classifying and predicting according to certain flood features. Based on fuzzy cluster iteration algorithm, history floods are classified into several specified groups, and by cluster validity evaluating, the optimal partition number and corresponding cluster centers may be obtained. Then taking the optimal cluster centers as the criterion of recognizing flood type and importing fuzzy pattern recognition theory, the real-time flood type may be predicted. Finally, the approach is applied to flood classifying and recognizing of Huanren reservoir basin and the results demonstrate the feasibility and practicability of this method.