Non‐linear mixed‐effects models for time series forecasting of smart meter demand

Abstract Buildings are typically equipped with smart meters to measure electricity demand at regular intervals. Smart meter data for a single building have many uses, such as forecasting and assessing overall performance. However, when available from multiple buildings, there additional applications that rarely explored. For instance, we can explore how different characteristics influence energy demand. If each is treated random effect handled fixed effects, mixed‐effects model be used estimate affect usage. In this paper, demonstrate producing 1‐day‐ahead predictions 123 commercial office buildings using mixed models improve accuracy. We experiment intercept, intercept slope non‐linear models. The predictive performance of the tested against naive, linear benchmark fitted separately. This research justifies accuracy quantify changes in consumption under configuration scenarios.