Development and application of hourly building cooling load prediction model

Abstract

In this paper, we employ support vector machine (SVM) and conventional artificial neural network to establish the prediction models of hourly cooling load in the building and the application cases in one office building and one library show that SVM method and conventional artificial neural network both can be effective for the prediction of hourly building cooling load. But comparing with conventional artificial neural network techniques, SVM can achieve better accuracy and generalization. It is a promising method to use SVM to predict the hourly cooling load in the building. Furthermore, a prediction software is developed for the on-line hourly building cooling load prediction based on the SVM method and artificial neural network.