Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

TrustEnergy: A Unified Framework for Accurate and Reliable User-level Energy Usage Prediction

Dahai Yu, Rongchao Xu, Dingyi Zhuang, Yuheng Bu, Shenhao Wang, Guang Wang

TL;DR

TrustEnergy addresses the dual challenge of accurate and reliable user-level energy usage prediction by jointly modeling micro household and macro region patterns with a memory-augmented spatiotemporal graph neural network (MASTGNN) and by providing distribution-agnostic, time-adaptive uncertainty via Sequential Conformalized Quantile Regression (SCQR). The framework reduces computational complexity through a memory-augmented parameter pool and time-aware embeddings, enabling scalable learning for large-scale electricity data. Empirical evaluation on a Florida utility dataset shows around 5.4% improvements in prediction accuracy and 5.7% improvements in uncertainty quantification over strong baselines, with demonstrated robustness under extreme weather and strong generalizability to datasets from New York and California. Overall, TrustEnergy offers a scalable approach for calibrated, region-aware energy prediction, with practical impact for grid management and emergency response.

Abstract

Energy usage prediction is important for various real-world applications, including grid management, infrastructure planning, and disaster response. Although a plethora of deep learning approaches have been proposed to perform this task, most of them either overlook the essential spatial correlations across households or fail to scale to individualized prediction, making them less effective for accurate fine-grained user-level prediction. In addition, due to the dynamic and uncertain nature of energy usage caused by various factors such as extreme weather events, quantifying uncertainty for reliable prediction is also significant, but it has not been fully explored in existing work. In this paper, we propose a unified framework called TrustEnergy for accurate and reliable user-level energy usage prediction. There are two key technical components in TrustEnergy, (i) a Hierarchical Spatiotemporal Representation module to efficiently capture both macro and micro energy usage patterns with a novel memory-augmented spatiotemporal graph neural network, and (ii) an innovative Sequential Conformalized Quantile Regression module to dynamically adjust uncertainty bounds to ensure valid prediction intervals over time, without making strong assumptions about the underlying data distribution. We implement and evaluate our TrustEnergy framework by working with an electricity provider in Florida, and the results show our TrustEnergy can achieve a 5.4% increase in prediction accuracy and 5.7% improvement in uncertainty quantification compared to state-of-the-art baselines.

TrustEnergy: A Unified Framework for Accurate and Reliable User-level Energy Usage Prediction

TL;DR

TrustEnergy addresses the dual challenge of accurate and reliable user-level energy usage prediction by jointly modeling micro household and macro region patterns with a memory-augmented spatiotemporal graph neural network (MASTGNN) and by providing distribution-agnostic, time-adaptive uncertainty via Sequential Conformalized Quantile Regression (SCQR). The framework reduces computational complexity through a memory-augmented parameter pool and time-aware embeddings, enabling scalable learning for large-scale electricity data. Empirical evaluation on a Florida utility dataset shows around 5.4% improvements in prediction accuracy and 5.7% improvements in uncertainty quantification over strong baselines, with demonstrated robustness under extreme weather and strong generalizability to datasets from New York and California. Overall, TrustEnergy offers a scalable approach for calibrated, region-aware energy prediction, with practical impact for grid management and emergency response.

Abstract

Energy usage prediction is important for various real-world applications, including grid management, infrastructure planning, and disaster response. Although a plethora of deep learning approaches have been proposed to perform this task, most of them either overlook the essential spatial correlations across households or fail to scale to individualized prediction, making them less effective for accurate fine-grained user-level prediction. In addition, due to the dynamic and uncertain nature of energy usage caused by various factors such as extreme weather events, quantifying uncertainty for reliable prediction is also significant, but it has not been fully explored in existing work. In this paper, we propose a unified framework called TrustEnergy for accurate and reliable user-level energy usage prediction. There are two key technical components in TrustEnergy, (i) a Hierarchical Spatiotemporal Representation module to efficiently capture both macro and micro energy usage patterns with a novel memory-augmented spatiotemporal graph neural network, and (ii) an innovative Sequential Conformalized Quantile Regression module to dynamically adjust uncertainty bounds to ensure valid prediction intervals over time, without making strong assumptions about the underlying data distribution. We implement and evaluate our TrustEnergy framework by working with an electricity provider in Florida, and the results show our TrustEnergy can achieve a 5.4% increase in prediction accuracy and 5.7% improvement in uncertainty quantification compared to state-of-the-art baselines.
Paper Structure (23 sections, 16 equations, 3 figures, 4 tables, 1 algorithm)

This paper contains 23 sections, 16 equations, 3 figures, 4 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Problem Formulation
  3. Hierarchical Graphs Construction
  4. Energy Usage Prediction
  5. Methodology
  6. Hierarchical Spatiotemporal Representation
  7. Parameter Pool Construction
  8. Parameter Updating Mechanism
  9. Loss Function
  10. Sequential Conformalized Quantile Regression
  11. Evaluation
  12. Experimental Setup
  13. Datasets
  14. Baselines
  15. Evaluation Metrics
  16. ...and 8 more sections

Figures (3)

  • Figure 1: An overall framework of TrustEnergy, which consists of (i) a Hierarchical Spatiotemporal Representation Learning module to efficiently capture both macro-level and micro-level patterns, and (ii) a Sequential Conformalized Quantile Regression module to dynamically adjust uncertainty bounds to ensure valid prediction intervals over time.
  • Figure 2: Uncertainty calibration results.
  • Figure 3: Prediction results under diverse extreme weather conditions.