On the Sunspot Time Series Prediction Using Jordon Elman Artificial Neural Network (ann)

In this paper, multi step ahead prediction of monthly sunspot real time series are carried out. This series is highly chaotic in nature [7]. This paper compares performance of proposed Jordan Elman Neural Network with TLRNN (Time lag recurrent neural network), and RNN (Recurrent neural network) for multi-step ahead (1, 6, 12, 18, 24) predictions. It is seen that the proposed neural network model clearly outperforms TLRNN(Time lag recurrent neural network), and RNN(Recurrent neural network) in various performance measures such as MSE (Mean square error), NMSE (Normalized mean square error) and r (correlation coefficient) on testing as well as training data set for 1,6,12,18,and 24 months ahead prediction of sunspot time series. Parameters are calculated by using software, “Neurosolution 5.0”. Neurosolution is an object oriented environment for designing, prototyping, simulating, and deploying artificial neural network (ANN) solutions [26].