Main Article Content
Forecasting the amount of inflow discharge is very necessary in the preparation of the operating pattern of water building so, as to maximize the function of the building is as a shelter to accommodate water for the purposes of hydropower, irrigation, raw water supply, ground water supply and tourism. Therefore it is necessary to find the right method to produce the best forecast with a small error by using Artificial Neural Network (ANN) method which is one of artificial intelligence. This method is the network of a group of small processing unit that are modelled based on human neural network consist of the input layer, the hidden layer and the output layer. In the training process, the data is divided into 2 parts, namely data to build the network (training) and data for forecasting (testing) and next done trial and error simulation to determine the many neurons needed on the hidden layer that produces the smallest Mean Square Error (MSE). In this research, Conjugate Gradient Fletcher-Reeves Updates algorithm is used to accelerate backpropagation training to obtain the best result consisting of 1 input layer, 2 hidden layers and 1 output layer or can be called architecture 1 - 2 - 1 with MSE value 0.2336 from October 2015 to February 2017 period.