An Optimal Neural Network for Hourly and Daily Energy Consumption Prediction in Buildings

In this work, hourly and daily energy consumption prediction has been carried out using multi-layer feed forward neural network. The network designed in the proposed architecture three layers, namely input layer, hidden output layer. layer had eight neurons, one neuron, number of neurons was varied to find an optimal for accurate prediction. Different parameters were repeatedly, accuracy observed each combination different optimized parameters. For prediction, a total six weeks data (September 1 October 12, 2004) 10 residential buildings used whereas 52 (January 2004 December 30 used. To evaluate performance approach, evaluation measurements applied.