Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Automated Machine Learning: A Case Study on Non-Intrusive Appliance Load Monitoring

Armin Moin, Ukrit Wattanavaekin, Alexandra Lungu, Stephan Rössler, Stephan Günnemann

TL;DR

This paper tackles Non-Intrusive Appliance Load Monitoring (NIALM) by introducing AutoML4NIALM, a Bayesian-Optimization-based AutoML framework that automatically selects ML models and hyper-parameters for disaggregating household energy signals. Through a fair benchmarking against state-of-the-art methods on the REDD dataset, the authors show that simple models like Decision Trees can outperform complex architectures in many scenarios, while AutoML can further improve regression and classification metrics across multiple models, including GRUs and FCNNs. An open-source tool is released to facilitate industry adoption, emphasizing practical guidance to prevent common data-splitting errors in time-series. The work highlights there is no one-size-fits-all solution for NIALM and suggests extending AutoML coverage, GUI support, and cross-domain applicability as future directions with potential for broader energy-management impact.

Abstract

We propose a novel approach to enable Automated Machine Learning (AutoML) for Non-Intrusive Appliance Load Monitoring (NIALM), also known as Energy Disaggregation, through Bayesian Optimization. NIALM offers a cost-effective alternative to smart meters for measuring the energy consumption of electric devices and appliances. NIALM methods analyze the entire power consumption signal of a household and predict the type of appliances as well as their individual power consumption (i.e., their contributions to the aggregated signal). We enable NIALM domain experts and practitioners who typically have no deep data analytics or Machine Learning (ML) skills to benefit from state-of-the-art ML approaches to NIALM. Further, we conduct a survey and benchmarking of the state of the art and show that in many cases, simple and basic ML models and algorithms, such as Decision Trees, outperform the state of the art. Finally, we present our open-source tool, AutoML4NIALM, which will facilitate the exploitation of existing methods for NIALM in the industry.

Automated Machine Learning: A Case Study on Non-Intrusive Appliance Load Monitoring

TL;DR

This paper tackles Non-Intrusive Appliance Load Monitoring (NIALM) by introducing AutoML4NIALM, a Bayesian-Optimization-based AutoML framework that automatically selects ML models and hyper-parameters for disaggregating household energy signals. Through a fair benchmarking against state-of-the-art methods on the REDD dataset, the authors show that simple models like Decision Trees can outperform complex architectures in many scenarios, while AutoML can further improve regression and classification metrics across multiple models, including GRUs and FCNNs. An open-source tool is released to facilitate industry adoption, emphasizing practical guidance to prevent common data-splitting errors in time-series. The work highlights there is no one-size-fits-all solution for NIALM and suggests extending AutoML coverage, GUI support, and cross-domain applicability as future directions with potential for broader energy-management impact.

Abstract

We propose a novel approach to enable Automated Machine Learning (AutoML) for Non-Intrusive Appliance Load Monitoring (NIALM), also known as Energy Disaggregation, through Bayesian Optimization. NIALM offers a cost-effective alternative to smart meters for measuring the energy consumption of electric devices and appliances. NIALM methods analyze the entire power consumption signal of a household and predict the type of appliances as well as their individual power consumption (i.e., their contributions to the aggregated signal). We enable NIALM domain experts and practitioners who typically have no deep data analytics or Machine Learning (ML) skills to benefit from state-of-the-art ML approaches to NIALM. Further, we conduct a survey and benchmarking of the state of the art and show that in many cases, simple and basic ML models and algorithms, such as Decision Trees, outperform the state of the art. Finally, we present our open-source tool, AutoML4NIALM, which will facilitate the exploitation of existing methods for NIALM in the industry.
Paper Structure (16 sections, 8 equations, 6 figures, 5 tables)

This paper contains 16 sections, 8 equations, 6 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Literature Survey & Benchmarking
  3. Proposed AutoML Approach
  4. Problem Formulation
  5. Automated ML for NIALM
  6. Experimental Results
  7. Evaluation Metrics
  8. Mean Absolute Error (MAE)
  9. Disaggregation Accuracy
  10. Classification Accuracy
  11. Precision, Recall and F1-Score
  12. Normalized Absolute Distance
  13. Experimental Setup
  14. Extended Benchmarking
  15. AutoML Experiments
  16. ...and 1 more sections

Figures (6)

  • Figure 1: Our proposed model for the search space of the AutoML approach
  • Figure 2: Benchmarking using Disaggregation Accuracy - color codes: dark blue, orange, gray, yellow and light blue represent fridge, lights, sockets, washer/dryer and average for all appliances, respectively.
  • Figure 3: Benchmarking using Classification Accuracy - color codes: dark blue, orange, gray, yellow and light blue represent fridge, lights, sockets, washer/dryer and average for all appliances, respectively.
  • Figure 4: Benchmarking using F1-Score - color codes: dark blue, orange, gray, yellow and light blue represent fridge, lights, sockets, washer/dryer and average for all appliances, respectively.
  • Figure 5: Benchmarking using Normalized Absolute Distance (NAD) - color codes: dark blue, orange, gray, yellow and light blue represent fridge, lights, sockets, washer/dryer and average for all appliances, respectively.
  • ...and 1 more figures