Artificial neural network for building energy consumption prediction

The optimization of energy consumption is important, especially in building these days. The energy consumption of Mechanical – Ventilating – Air conditioning (MVAC) systems always the largest proportion in total energy consumption. One of the most effective approach is the estimation and prediction as close as possible to the real operation mode of the MVAC in reality. However, the tasks of energy consumption estimation request large efforts, time consuming but low precise results. Most designers roughly calculate the energy consumption at first and the using automation control to manipulate the system operation later. With the help of the newest, cutting edge technology in soft computing like ANN, ANFIS and intelligent optimization methods, the designers now can apply these technologies to achieve the reliable, rapid, precise results for the further convincing designs. In this paper, the authors simply apply the Multi Layers, Back Propaganda Neural Network (MLB ANN) to estimate the building Heating Load (HL) and Cooling Load (CL). The Mean Square Errors (MSE) and Regression R Values (R) are used as the indicators for ANN with the dataset provided open access by UCI. The results prove that ANN can be reliable approach to predict energy consumption for complex, multi design option building. Keywords: ANN, Sustainable Design, Artificial Intelligence