Noise Injection Free Privacy Preserving Energy Consumption Profiling

The application of data-driven science to power systems is rapidly growing, particularly in consumer behavior analysis. However, accessing user electricity consumption data has raised public concerns about privacy breaches. Classical privacy preserving mechanisms tend to inject noise into data to protect user privacy, but this usually destroys the data structure. This paper identifies internal data noise to reduce the need for external noise injection, protecting the structure of the data. The study uses the Rényi Differential Privacy framework and a comprehensive numerical analysis to quantify the relationship between privacy protection levels and synthetic data’s solar energy proportion, which could serve as a guideline to both system operators and consumers when making privacy preserving decisions.