Abnormality detection by model-based estimation of power consumption

Heating, ventilation, and air conditioning (HVAC) systems occupy a large amount of power consumption in buildings. We introduce a case study of power consumption in the research and teaching building at National Taiwan University in July, 2012. Power consumption, the parameters of the HVAC system, the number of occupants, and the climate information are collected for power related analysis. We proposed an approximation for minimum cooling demands to observe the appropriateness of power consumption. The results show that the real cooling supply is much higher than the minimum cooling demands which would be a chance to mitigate the power consumption. In addition, we also investigate to detect events for preventing the potential errors. It is useful for promoting efficiency for power usage through the analysis of events. The results show that some of events could be detected by our methods, but several events are difficult to explain. We will improve the methods for abnormal detection and collect more data with variant events in the future.