What question did this study set out to answer?

The aim is to develop an integrated framework for energy load forecasting and reliable anomaly diagnosis in smart buildings.

April 17, 2026Applied SciencesOpen Access

Explainable Smart-Building Energy Consumption Forecasting and Anomaly Diagnosis Framework Based on Multi-Head Transformer and Dual-Stream Detection

View Full Paper

Ask AI

Bookmark

Authors

Integrated framework demonstrates accurate load forecasting and anomaly detection in smart buildings, suggesting a data-driven monitoring solution.

The aim is to develop an integrated framework for energy load forecasting and reliable anomaly diagnosis in smart buildings.
Developed a Transformer-based sequence-to-sequence model for hourly energy demand forecasting.
Implemented a dual-scale strategy for anomaly diagnosis, focusing on both short-term and long-term anomalies.
Utilized adaptive thresholds to detect short-term anomalies from forecasting residuals.
Compared current residual patterns with historical baselines for long-term anomaly detection.
Provided SHAP-based explanations for interpreting predictions and detected anomalies.
Achieved a mean absolute percentage error of approximately 3% in forecasting performance.
Demonstrated superior performance against conventional baselines and recent Transformer-based models.
Effectively distinguished various types of anomalies: point, pattern, and composite.

Cai et al. (2026) studied this question.

View Full Paper

Ask AI

Bookmark

April 17, 2026Applied SciencesOpen Access

View Full Paper

Ask AI

Bookmark

Authors

Integrated framework demonstrates accurate load forecasting and anomaly detection in smart buildings, suggesting a data-driven monitoring solution.

The aim is to develop an integrated framework for energy load forecasting and reliable anomaly diagnosis in smart buildings.
Developed a Transformer-based sequence-to-sequence model for hourly energy demand forecasting.
Implemented a dual-scale strategy for anomaly diagnosis, focusing on both short-term and long-term anomalies.
Utilized adaptive thresholds to detect short-term anomalies from forecasting residuals.
Compared current residual patterns with historical baselines for long-term anomaly detection.
Provided SHAP-based explanations for interpreting predictions and detected anomalies.
Achieved a mean absolute percentage error of approximately 3% in forecasting performance.
Demonstrated superior performance against conventional baselines and recent Transformer-based models.
Effectively distinguished various types of anomalies: point, pattern, and composite.

Cai et al. (2026) studied this question.

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

View Full Paper

Ask AI

Bookmark