What question did this study set out to answer?

The aim is to develop a health-aware Non-Intrusive Load Monitoring system to identify appliance energy usage and detect faults.

May 9, 20260 citationsOpen Access

View Full Paper

Health-Aware Non-Intrusive Load Monitoring Using Dual-Branch Attention Network and Variance-Anchored XGBoost Classifier

AAAnkit Ambasta

Key Points

The aim is to develop a health-aware Non-Intrusive Load Monitoring system to identify appliance energy usage and detect faults.
Combines a dual-branch attention-based deep learning model for energy disaggregation
Integrates an XGBoost-based model for anomaly detection
Experiments conducted using the REFIT dataset
Improved performance compared to CNN, LSTM, GRU, DTW, and Random Forest baselines
Enhanced accuracy in identifying energy consumption across appliances
Effective fault detection capabilities noted in test scenarios

Abstract

This paper presents a health-aware Non-Intrusive Load Monitoring (NILM) system that can both identify household appliance energy usage and detect possible appliance faults using only smart meter data. The proposed framework combines a dual-branch attention-based deep learning model for energy disaggregation with an XGBoost-based anomaly detection model for appliance health monitoring. Experiments on the REFIT dataset show improved performance over existing CNN, LSTM, GRU, DTW, and Random Forest baselines across multiple household appliances.

Demander à l'IA

Bookmark

View Full Paper

Demander à l'IA

Bookmark

View Full Paper

Health-Aware Non-Intrusive Load Monitoring Using Dual-Branch Attention Network and Variance-Anchored XGBoost Classifier

Key Points

Abstract

Cite This Study