Comparing Methods of Artificial Neural Network and Fuzzy System in Determining Pre-flooding Warning (Case Study: Zard River Sub-basin- Khuzestan Province)

Abstract
One method of flood forecasting and flood control in rivers is ‘flood routing'. The relationship between precipitation and runoff and creating flooding in the region is not linear mathematical relationship which we can predict flooding in one region and such phenomena should be regarded as a model. Artificial intelligence methods such as artificial neural network and fuzzy inference system can be used as a good method in this field. In this study, using artificial neural network and fuzzy inference system, which are two types of the most widely used computational intelligence, we attempt to predict flood in Zard River. For the implementation both methods, first, the necessary data were collected and then wrong data were excluded from the data set and the data have been normalized. Modeling using artificial neural networks using MATLAB software coding was performed on data. To implement, the fuzzy inference system were used from prepared data. In this study, types of artificial neural networks structures with different number of neurons and hidden layers, number of educational courses and different functions have been performed on the data until obtaining the best structure for study area. Fuzzy inference models were implemented until the best model is chosen. Results showed that in general, fuzzy inference system have a better simulate data in the studied area and better and more accurate results than the artificial neural network model is showed. Also, values of MSE and r in fuzzy inference system and artificial neural network is equal to 0.2196, 0.0297, 0.7667 and 0.96 respectively which shows higher accuracy of fuzzy inference system for predicting floods in the our area of the study.