<![CDATA[Artificial Neural Network (ANN) is a technology that was developed based on the principle of biological neural networks in humans, can betrained to predict what will happen future based on patterns occurrence existing in the past. ANN has the ability to remember and make ageneralization of what has happened before. Artificial neural networks can train the network to get the balance between the ability of the networkto recognize patterns (historical data) are used for training as well as the network's ability to respond correctly to the input patterns are similar (butnot the same) to the pattern used during training. The purpose of this study was to determine the best number of input patterns using neuralnetworks, back propagation architecture.This research method using quantitative descriptive methods with techniques of data collection sources or agencies related to the research data usedare secondary data. Stages of the research carried out by preparing the data discharge in the year 2001-2012. For the simulation of discharge datausing Artificial Neural Networks (ANN) backpropagation with the help of MATLAB software.The results showed that the number of input pattern is best with the input pattern data input discharge 8 years. The best simulation results aresimulated discharge data with the data input discharge 8 years with the data output discharge 8 years. Reliability of simulation results only reaches64.68%, the simulation results have a fairly good result level of reliability and 95 % Confidence qualify, but the parameter of the model need to bemodified to apply to apply to other watersheds.]]>
Copyrights © 2014
