Current studies conducted by JICA, AMPRI and IFC-World Bank, reported that large commercial buildings in Indonesia are not energy and water efficient. One of the cause is the lack of regulation. Meanwhile, effective regulations to reduce energy and water consumption are the concern mostly in a new building to obtain a building permit. This strategy is understandable as retrofitting existing buildings are often more difficult to be implemented, and enforcement is still a major issue in Indonesia. Local governments are currently working on streamlining building permit process as well as developing an online monitoring system for existing buildings. By applying a Building Energy Management System (BEMS) enables to reduce energy consumption up to 15%. An energy monitoring system was designed and installed through this research for Department of Nuclear Engineering and Engineering Physics (DNEEP) building, Faculty of Engineering, Universitas Gadjah Mada. It is a 20 years old two-story building used for educational activities, which consist of classrooms, laboratories, offices and storage spaces. An audit energy was done recently in 2015 where an energy consumption of 261.299,636 kWh/year.m2 was reported. In the existing condition, a power meter is inaccessible and therefore, the only feedback of occupancy behavior in the energy consumption is through the electricity bill. The previous study has shown that building occupants would behave more efficiently if the amount of energy used is notified, and the amount of energy savings are recorded. However, these energy monitoring systems are considered expensive and uniquely tailored for every building. This research aims to design and install a cost effective BEMS based on occupant’s satisfactory assessment of the lighting, acoustics, and air conditioning quality. The data will be used as a decision supporting system (DSS) by building management through the use of a GUI. The design of the interface was based on a survey result from the prospective users. Installed energy monitoring system uses a current sensor with an accuracy of 98% and a precision of 0.04 A while the voltage sensor with an accuracy of 98% and a precision of 0.58 V. The performance testing shows that the number of web clients influences delay of data transmission. The result of the survey shows that GUI is categorized as fair in design without a significant difference between the perceptions of users with and without survey supervision.
Machine Learning in Smart Home Energy Monitoring System
