Handling Data Heterogeneity in Electricity Load Disaggregation via Optimized Complete Ensemble Empirical Mode...

Area of Science:

• Energy research
• Artificial intelligence
• Environmental science

Background:

• Global warming necessitates carbon emission reduction.
• Smart meters collect household electricity usage data.
• Electricity load disaggregation (ELD) analyzes individual appliance consumption.

Purpose of the Study:

• To propose a novel powerline noise transformation approach for merging ELD datasets.
• To enhance the size and utility of training datasets for ELD models.
• To improve the accuracy and reliability of electricity usage analysis.

Main Methods:

• Developed an optimized complete ensemble empirical model decomposition and wavelet packet transform (OCEEMD-WPT) method.
• Applied the OCEEMD-WPT approach to merge multiple ELD datasets.
• Compared the proposed method against CEEMD-WPT, CEEMD, and WPT.

Main Results:

• The OCEEMD-WPT approach significantly improved the signal-to-noise ratio (SNR).
• Merging datasets increased training data size and facilitated cross-dataset utilization.
• The method demonstrated superior performance compared to existing techniques.

Conclusions:

• The proposed OCEEMD-WPT method is effective for merging ELD datasets.
• This approach offers practical benefits for energy conservation efforts.
• Enhanced ELD analysis contributes to a better understanding of electricity consumption patterns.