DESIGN OF PREDICTION SYSTEM USED TO PREDICT BUILDING’S ELECTRICITY USAGE PROFILE

• Main Author: Wulandari Setiadi (13314009), Billy Suyapmo (13314064) , Inneke
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/28066
• Institution: Institut Teknologi Bandung

<p align="justify">Monitoring building's electricity usage is part of energy management which is essential in operational planning or budgeting. For these purposes, the understanding of energy usage baseline in the future is needed to be consider in taking further action both in operational or budget. By using the unique characteristic of each bulding’s electricity consumption then the regression model of the electricity consumption can be generated using Support Vector Regression. Each characteristics differ from its usage, type of the bulding, occupant and many more. The ability of Support Vector Regression in finding the relation of each data by transferring the data to higher dimension by using kernel tricks. Only in this higher dimension that the relation of each data can be found and the regression model can be created. This model therefore used to predict electricity usage per hour of TP Rachmat ITB (Labtek VI) Building. From time series analysis in Support Vector Regression, the ouput of prediction system is in the range of MAPE 8,49% until 12,33% and RMSE 2,42% until 6,21%. <p align="justify"> <br />