Event-Induced Anomalies in Energy Consumption Using the Wavelet Transform and Long Short-Term Memory Neural Network: Case Study at Arizona State University

This research investigates the complex relationship between events and anomalies in energy consumption patterns by employing wavelet transform techniques and developing long short-term memory (LSTM) neural network. The study seeks to understand and predict energy consumption anomalies based on events with high transient occupancy. The results of this study are significant to all industry sectors, particularly in energy consumption metrics for utility companies and property managers. Utility companies can use such advancements to optimize energy distribution more efficiently, aligning supply with actual demand to minimize waste and reduce carbon footprint. Similarly, property managers can use these models to manage energy consumption in buildings more effectively, thereby contributing to broader sustainability efforts. This paper’s application of wavelet transform and LSTM models to study event-induced anomalies enhances the accuracy of energy consumption predictions by addressing the relationship between transient occupancy and energy consumption, addressing energy oversupply concerns, and supporting environmental conservation efforts by reducing unnecessary energy consumption.