Predicting plug loads with occupant count data through a deep learning approach

Abstract

Predictive control has gained increasing attention for its ability to reduce energy consumption and improve occupant comfort in buildings. The plug loads prediction is a key component for the predictive building controls, as plug loads is a major source of internal heat gains in buildings. This study proposed a novel method to apply the Long-Short-Term-Memory (LSTM) Network, a special form of Recurrent Neural Network, to predict plug loads. The occupant count and the time have been confirmed to drive the plug load profile and thus selected as the features for the plug load prediction. The LSTM network was trained and tested with ground truth occupant count data collected from a real office building in Berkeley, California. Results from the LSTM network markedly improve the prediction accuracy compared with traditional linear regression methods and the classical Artificial Neural Network. 95% of 1-hour predictions from LSTM network are within ±1 kW of the actual plug loads, given the average plug loads during the office hour is 8.6 kW. The CV(RMSE) of the predicted plug load is 11% for the next hour, and 20% for the next eight hours. Lastly, we compared four prediction approaches with the office building we monitored: LSTM vs. ARIMA, with occupant counts vs. without occupant counts. It was found, the prediction error of the LSTM approach is around 4% less than the ARIMA approach. Using occupant counts as an exogenous input could further reduce the prediction error by 5%-6%. The findings of this paper could shed light on the plug load prediction for building control optimizations such as model-predictive control.

The effect of occupant distribution on energy consumption and COVID-19 infection in buildings: A case study of university building

“…In order to find the best population density in a building, Kang et al [ 20 ] studied the effect of occupant density on 105 buildings and derived a direct relationship between the number of occupants and the building's energy consumption. Li and Yao [ 21 ], and Want et al [ 22 ], considering the occupants, implemented machine learning techniques in modeling and predicting the energy consumption of buildings and concluded that occupant's presence and behaviors are one of the most determinant factors in energy consumption of buildings. Sun et al [ 23 ] proposed a framework for data-driven occupant behavior analytics in residential buildings and concluded that occupant's effect on a building's energy consumption is significant, whether it's a residential or a commercial building.…”

A framework for estimating the energy-saving potential of occupant behaviour improvement

“…The rest plug load is strongly influenced by the occupant count in the zone and status of appliances when they are not engaged (namely standby status). According to previous studies by Mahdavi et al [41] and Wang et al [42], the plug load shows a linear relationship to the occupancy fraction. Thus, in this study, the plug load fraction of a zone is determined by the linear function of the occupancy fraction as illustrated in Figure 7.…”

Building thermal load prediction through shallow machine learning and deep learning

“…Fan et al compared seven machine learning algorithms (multiple linear regression, elastic net, random forests, gradient boosting machines, SVM, extreme gradient boosting, and deep neural network) and found extreme gradient boosting combined with deep auto-coding performed best [28]. Wang et al compared LSTM and ARIMA and found LSTM performs better than ARIMA in plug load prediction [29]. Rather than selecting the best predictor, ensembling multiple different predictors into one model could provide better generalization performance [23].…”