OPTIMIZATION OF NEURAL NETWORK'S TRAINING SETS VIA CLUSTERING: APPLICATION IN SOLAR COLLECTOR REPRESENTATION

摘要

Due to the necessity of new ways of energy producing, solar collector systems have been widely used around the world. There are mathematical models that calculate the efficiency of those systems; however these models involve several parameters that may lead to nonlinear equations of the process. Artificial Neural Networks have been proposed in this work as an alternative of those models. However, a better modeling of the process by means of ANN depends on a representative training set; thus, in order to better define the training set, the clustering technique called k-means has been used in this work.