Short-term HVAC load forecasting algorithms for home energy management

Abstract

This paper presents the forecasting algorithms for determining the electricity usage and operation status of residential heating, ventilating, and air conditioning (HVAC) systems. Two algorithms are presented based on what types of measured data can be received by the home energy management system (HEMS). Algorithm 1 is developed assuming only HVAC status is available to forecast the future HVAC usage. Algorithm 2 is developed for cases that the HVAC operation status, room temperature and outdoor temperature time series are known. The sensitivity of the room temperature change rate to outdoor temperature is derived and used to forecast the HVAC operation status. Results show that Algorithm 1 performs well for very short term forecast (less than 1 hour) and Algorithm 2 outperforms Algorithm 1 when forecasting HVAC behaviors for longer periods (from one hour to several hours) under a broader operation conditions such as continuously running or cold start. Both algorithms are measurement-based and require little computational resources and time to implement so that they fit well for providing HVAC status estimation to the HEM system for scheduling HVAC loads.