Reproducible Non-Intrusive Load Monitoring: A Survey of Open-Source Methods and Practical Challenges

Non-intrusive load monitoring (NILM) is a promising solution for improving energy efficiency and promoting sustainable electricity usage by disaggregating appliance-level consumption from aggregate smart meter data. However, despite extensive research, many proposed solutions still lack detailed implementation resources, such as data and code, making reproducibility challenging for both researchers and industry practitioners. This paper presents a comprehensive review of open-source NILM approaches, with a specific focus on reproducibility. We survey both high- and low-frequency NILM methods, analyze publicly available implementations, and conduct reproducibility tests to assess alignment between reported and reproduced results. Through this analysis, we identify key challenges related to dataset inconsistencies, evaluation metrics, and experimental settings. We also discuss strategies for improving methodological transparency, comparability, and reliability of NILM research. Finally, we explore the practical challenges and opportunities for deploying NILM in real-world applications.