Abstract:To meet energy-reduction goals, cities are challenged with assessing building energy performance and prioritizing efficiency upgrades across existing buildings. Although current top-down building energy benchmarking approaches are useful for identifying overall efficient and poor performers across a portfolio of buildings at a city scale, they are limited in their ability to provide actionable insights regarding efficiency opportunities. Concurrently, advances in smart metering data analytics combined with new data streams available via smart metering infrastructure present the opportunity to incorporate previously undetectable temporal fluctuations into top-down building benchmarking analyses. This paper leveraged smart meter electricity data to develop daily building energy benchmarks segmented by strategic periods to quantify their variation from conventional, annual energy benchmarking strategies and investigate how such metrics can lead to near real-time energy management. The periods considered include occupied periods during the school year, unoccupied periods during the school year, occupied periods during the summer, unoccupied periods during the summer, and peak summer demand periods. Results showed that temporally segmented building energy benchmarks are distinct from a building’s overall benchmark. This demonstrates that a building’s overall benchmark masks periods in which a building is over- or underperforming during the day, week, or month; thus, temporally segmented energy benchmarks can provide a more specific and accurate measure for building efficiency. We discussed how these findings establish the foundation for digital twin–enabled urban energy management platforms by enabling identification of building retrofit strategies and near-real-time efficiency in the context of the performance of an entire building portfolio. Temporally segmented energy benchmarking measures generated from smart meter data streams are a critical step for integrating smart meter analytics with building energy benchmarking techniques, and for conducting smarter energy management across a large geographic scale of buildings.

Data-Driven Building Load Profiling and Energy Management

Wide Area Electric Load Modeling Based on Smart Meter

Electric Load Data Characterising and Forecasting Based on Trend Index and Auto-Encoders

A Weekly Load Data Mining Approach Based on Hidden Markov Model

Case studies of fault diagnosis and energy saving in buildings using data mining techniques

Advanced Data Analytics for Enhancing Building Performances: from Data-Driven to Big Data-Driven Approaches

A review of data mining technologies in building energy systems: Load prediction, pattern identification, fault detection and diagnosis

A review of data-driven approaches for prediction and classification of building energy consumption

Novel approach to energy consumption estimation in smart homes: application of data mining and optimization techniques

Data-Driven Evaluation of Building Demand Response Capacity

Data-driven approaches to achieve energy transition: Statistical analysis of demand-side management and smart decision making in urban energy hub networks

Statistical characterization of electricity use profile: Leveraging data analytics for stochastic simulation in a smart campus

Smart City Digital Twin–Enabled Energy Management: Toward Real-Time Urban Building Energy Benchmarking

Building Categorization Revisited: A Clustering-Based Approach to Using Smart Meter Data for Building Energy Benchmarking

Data Driven Approaches for Prediction of Building Energy Consumption at Urban Level

A Framework for the Automatic Integration and Diagnosis of Building Energy Consumption Data

Data-driven energy management in residential areas leveraging demand response

Predicting the electric power consumption of office buildings based on dynamic and static hybrid data analysis

A hybrid evolutionary and machine learning approach for smart building: Sustainable building energy management design

Power Management in Smart Residential Building with Deep Learning Model for Occupancy Detection by Usage Pattern of Electric Appliances

Building Electricity Consumption Analysis: Data-Driven Approach with Preprocessing, Visualization, and Cluster Analysis