Prediction of building lighting energy consumption based on support vector regression

• Published in: 2013 9th Asian Control Conference (ASCC) pp. 1 - 5
• Main Authors: Dandan Liu, Qijun Chen
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.06.2013
• Subjects: Artificial neural networks; building; Buildings; Energy consumption; Lighting; lighting energy consumption; prediction; Predictive models; Support vector machines; support vector regression; Training
• ISBN: 9781467357678; 1467357677
• DOI: 10.1109/ASCC.2013.6606376

Prediction of energy consumption is an important task in energy conservation. Due to support vector regression has good performance in dealing with non-linear data regression problem, in recent years it often was used to predict building energy consumption. Based on the historical data we conclude the relationship between lighting energy consumption and its influencing factors is non-linear. To develop accurate prediction model of lighting energy consumption, the support vector regression with radial basis function was applied. The forecast results indicate that the prediction accuracy of support vector regression is higher than neural networks. The prediction model can forecast the building hourly energy consumption and assess the impact of office building energy management plans.