Multilayer feed forward neural network to predict the speed of wind

An artificial neural network is massively interconnected network of neurons with parallel processing capability. It can learn from known instances of problem and can predict or solve unknown instances of problem. A nerual network is usually used to solve problem with non-linear distribution. In this paper, a neural network models are designed with varying parameters for real time datasets, for better efficiency. The neural network primarily contains three layers contains three layers: Input, Output and Hidden. Back propagation algorithm with feed forward is used to train the network with real time dataset. Random weights are assigned initially and the network is trained. The wind speed is predicted using various parameters. The error in the given network is calculated by using the target output, as in the case of supervised learning for a Neural Network. The network with the best fit is chosen, trained, validated and the prediction is made about the wind speed.