Energy Consumer Model Estimation Using Machine Learning

Abstract

A novel utility model of the daily electricity consumption patterns of any residential unit is proposed. The model is semi-parametric—its internal parameters are tractable to straightforward interpretations from the standpoints of microeconomics as well as game theory. The model considers fixed and shiftable loads. A modified Cobb–Douglas preference function is applied to incorporate utility tradeoffs in operating shiftable loads at different hours of the day based on the consumer’s intrinsic preference and time-of-use pricing. Model parameter estimation is formulated as a quadratic optimization problem with inequality constraints. The existence of a global optimum that is unique except under a very specific circumstance is theoretically established. A dual-gradient-ascent learning algorithm was developed to train the parameters that best fit real time-series load data. For units with single-point monitoring, an optional unsupervised learning algorithm for partial load disaggregation was developed. As a case study, the proposed model was trained with load profiles of two real-world residential units. The trained model is able to faithfully reproduce typical electricity consumption characteristics. Model interpretability was demonstrated by numerically evaluating the utility and cost functions with varying loads. These results illustrate the effectiveness of the proposed methodology.