%0 Journal Article
%T Forecasting of the Alavian Dam Inflow water Using Optimized Adaptive Neuro-Fuzzy Inference System (OANFIS)
%J Iranian Journal of Soil and Water Research
%I University of Tehran
%Z 2008-479X
%A Misaghi, Farhad
%D 2016
%\ 10/22/2016
%V 47
%N 3
%P 439-448
%! Forecasting of the Alavian Dam Inflow water Using Optimized Adaptive Neuro-Fuzzy Inference System (OANFIS)
%K forecasting
%K OANFIS
%K Sequential
%K Exhaustive Search Algorithms
%R 10.22059/ijswr.2016.59314
%X In this study, Optimized Adaptive Neuro-Fuzzy Inference System (OANFIS) was employed on a set of daily, weekly, 10-days and monthly data of inflow water into the Alavian Dam to predict the real-time inflow of the reservoir. Sequential and exhaustive search algorithms were used to determine the numbers and time steps of the model inputs and also reducing the predictionâ€™s errors. In sequential search stage, several inputs series in daily, weekly, 10 days and monthly scales were developed as inputs and those were compared with outflows in time t as expressed by V (t). Also in exhaustive search phase, combinations of 2 from 10 and 3 from 10 which was included 45 and 120 models of time scale of V (t-1) to V (t-10) as inputs were developed and compared with outputs in time t as Vt. Statistical techniques including goodness of fit was used to evaluate the developed models performance. In sequential algorithm with daily scale, in the first step the input of V (t-1) with RSME=0.211 MCM, in the second step the input combination of V (t-1) to V (t-8) with RSME=0.187 MCM and also in the third step V (t-1), V (t-3) and V (t-4) with RSME=1.525 MCM were selected. Also in weekly scale, in the first step the input of V (t-1) with RSME=0.175 MCM, in the second step the input combination of V (t-1) to V (t-8) with RSME=0.192 MCM and also in the third step V (t-1), V (t-3) and V (t-4) with RSME=0.391 MCM were selected. In all of the optimized models of the studied time steps, the inputs of the V(t-1) was recognized as an effective factor and models outputs were sensitive to this variable at this time step which had the least time difference with output.Â
%U https://ijswr.ut.ac.ir/article_59314_b9c2300cd6875541add6c581e8c3ceae.pdf