Abstract
For the average electricity consumer, electricity is viewed as an invisible commodity. The consumer is not aware of their energy usage or how fast the dollars are adding up until the electricity bill arrives a month later. In the traditional utility environment, there are no mechanisms in place to alter the consumers’ consumption patterns. However, with the adoption of the smart grid, consumers are given the means to track and adjust their energy usage to achieve demand response.
Demand response has been used frequently within the commercial and industrial sectors and given the emerging smart grid paradigm, it is now being expanded into the residential sector. As such, it not only promotes cost savings for the consumer, but also reduces the need for utility company generation capacity to supply peak time energy usage. Within this dissertation, the residential sector is investigated to identify gaps that need to be addressed to achieve demand response. As an integral component to help the consumer achieve demand response, a set of models are developed to predict the interior conditions within a building and provide information to the consumer identifying which loads impact the interior conditions to assist in planning of energy consumption. In particular, a set of predictive thermal response models can be embedded in an intelligent system framework which can then proffer decisions that lead to economic and environmental advantages to the consumer.
Our methodology is based on developing Autoregressive Integrated Moving Average (ARIMA) time series thermal response models from past observational data collected from a residence (i.e., smart home). A year of collected data is used for model development and analysis. Due to the nature of the time series and the objective of this analysis, 70% of the observations are used to train the models; the remaining 30% is used to test the models for forecasting accuracy.
The research shows that the ARIMA time series method can be used to obtain short-term (e.g., up to 6 hours) thermal response forecasts to not only provide consumers with insight on their comfort levels but also assist in identifying the contributing loads (e.g., heating, ventilation, and air-conditioning (HVAC), water heater) that consume energy and influence interior conditions. The advantage of this research is the ability to forecast interior conditions and identify contributing loads based on activity recognition conducted in the residence during varying outside temperature and time of day conditions. Validation of the forecasting models has been carried out by comparing the models’ output with the actual data collected from the interior temperature conditions.
