A Validation Study of Remote Energy Diagnosis Algorithm Performance through Actual Building Energy Data Analysis

기존 건물 에너지 데이터 분석을 통한 원격 에너지진단 알고리즘 성능 검증 연구

Energy reduction and efficiency in existing buildings are an essential part of energy saving in building sector. Measurement of energy consumption by end-use and analysing data to find opportunities in improvement of energy performance are the key steps and the most important part of energy saving activities for buildings. This paper introduces and studies a recently-developed remote energy diagnosis program which has a new approach in building energy audit and analysis processes. The program consists of two algorithms - disaggregation and prediction. The first one is an algorithm for disaggregation of electricity consumption by end-use which divides the actual whole electricity consumption data of a building into three different end-use loads - heating & cooling load, lighting & other power supply load, and base load. Building characteristic factors in both heating and cooling periods are calculated for each building in this algorithm. The second algorithm provides the prediction of weather-corrected electricity consumption of a whole building. The output of the first algorithm and present outdoor temperature are combined for the prediction of present normal electricity consumption. This program does not require any difficult-to-get information or any hardwares for measurement. The analysis procedure is a lot quicker than the traditional on-site diagnosis approach. The validation of this remote energy diagnosis program is made based on comparison with actual building electricity consumption data. The result shows that the mean bias error(MBE) is -0.12% and coefficient of variation of the root mean square error(CV) is 5.7%, which are much better results than ASHRAE standards of acceptable calibration tolerances of ${\pm}5%$ and 30% respectively.