Abstract:Understanding and modeling building electricity consumption is paramount for demand-side decarbonization. Using a data-driven approach, this study proposes a hierarchical model to generate annual hourly electricity consumption profiles for commercial buildings, to facilitate energy system design and renewable energy integration. Electricity consumption profiles and the corresponding weather data from 55 commercial buildings across all five climate zones in China were collected to address the scarcity of relevant open datasets. Furthermore, periodicity and correlation analyses of the collected dataset indicate that the annual electricity consumption profiles can be decomposed into daily fluctuations primarily influenced by climate change and intraday fluctuations, which are predominantly affected by building schedules. Based on these features, suitable algorithms were selected to address the issues independently in these two temporal paradigms, forming a hierarchical model. A regression tree was used to reveal the relationships between daily consumption and weather data as well as the climate zone label. Additionally, k-means clustering was employed to extract typical intra-day consumption patterns, followed by the deployment of a classification tree to predict intra-day pattern labels under specific conditions. In the cross-validation, the proposed hierarchical model achieved a high coefficient of determination (R2) of 0.941, which further improved to 0.949 after incorporating residual regression, demonstrating its high accuracy. Compared to the single-step gradient boosting decision tree (GBDT) model (R2 = 0.951), the proposed hierarchical model exhibited comparable accuracy and offered better interpretability and lower overall complexity, making it a favorable choice for engineering applications. This study provides insights into the development of a data-driven model for building electricity consumption profiles.

Disaggregating Power Consumption of Commercial Buildings Based on the Finite Mixture Model

A Generic Energy Disaggregation Approach: What and When Electrical Appliances are Used.

Interleaved Factorial Non-Homogeneous Hidden Markov Models for Energy Disaggregation

Energy Disaggregation for Real-Time Building Flexibility Detection

Non-Intrusive Energy Use Monitoring for a Group of Electrical Appliances

Identifying ZIP Coefficients of Aggregated Residential Load Model Using AMI Data

Feature Extraction and an Interpretable Hierarchical Model for Annual Hourly Electricity Consumption Profile of Commercial Buildings in China

A Dynamical Systems Approach to Energy Disaggregation

Modeling Energy Consumption in Residential Buildings: A Bottom-Up Analysis Based on Occupant Behavior Pattern Clustering and Stochastic Simulation

Low Complexity Energy Disaggregation Algorithm for Non-intrusive Load Monitoring

Grouping Of Dynamic Electricity Consumption Behaviour: An F-Divergence Based Hierarchical Clustering Model

Appliance Use Behavior Modelling and Evaluation in Residential Buildings: A Case Study of Television Energy Use

Energy consumption disaggregation in commercial buildings: a time series decomposition approach

A Data-Driven Machine Learning Approach for Consumer Modeling with Load Disaggregation

Plug and Play! A Simple, Universal Model for Energy Disaggregation

Activity Detection And Modeling Using Smart Meter Data: Concept And Case Studies

CAPED: Context-Aware Powerlet-Based Energy Disaggregation.

Adaptive modeling for Non-Intrusive Load Monitoring

A secure edge monitoring approach to unsupervised energy disaggregation using mean shift algorithm in residential buildings

PHASED: Phase-Aware Submodularity-Based Energy Disaggregation

Data-Driven Energy Consumption Modeling and Simulation Method of Discrete Manufacturing Workshop Based on Simplified Power Curve