توصيفگر ها :
شبكه عصبي پيش خور , روش ستاره منحني , شبكه آبسنجي , مدل رياضي نرون , قوانين يادگيري , قانون هاپفيلد، هب، دلتا، كاهش گراديان، كوهنن , پرستپرون , رودخانه بختياري , سيگمويد , تانژانت هيپربوليك , رواناب
چكيده انگليسي :
Abstract Detennination of rainfall and runoff relationship is one of the most important problems for hydrologists and engineers Rainfall runoff information is necessary for hydrologic designs and management purposes in a watershed This relationship is known to be highlyt nonlinear and complex The runoff is not only dependent on rainfall amount but also it relies on numerous factors such as initial soil moisture Jand use watershed geomorphology evaporation infiltration distribution duratipn of the rainfall and so on Although many watersheds have been gauged to provide continuous records of stream flow engineers are often faced with situations where Jittle or no information is available In such situation it is preferred to implement a simple black box model to identify a direct mapping between the input and output without paying attention to the detailed intemalF structure of the physica1 process Artificial neura1 network is capab1eto identify the complex non linear relationships between input and output data without any need to understandthe nature of the phenomena In this study back propagation neural network BPNN models were used to forecast daily river flows in Bakhtiyari basin The rainfall data from upstream stations was used for model investigation Two different activation functions Sigmoid and Tangent Hyperbolic were implemented in multi layer perceptron network MLP Due to probability nature of choosing effective inputs in ANN a statistical approach PACF was used to estimate the effective delays in the target stations The sensitivity analysis approach was used to investigate how the selected stations affect the result of modeling The result derived from ANN shows that the input pattern included rainfall data for the interested day and discharge data of a day before interested day is the best pattern input Finally the result derived by ANN was compared with data derived from SCS approach and new approach for estimation of CN was developed i li f l II II I J
