Privacy-preserving methods for smart-meter-based network simulations

TL;DR

This paper introduces two novel privacy-preserving data anonymisation methods for smart-meter data, enabling network simulation and analysis while complying with privacy regulations.

Contribution

It proposes two innovative anonymisation techniques for smart-meter data that maintain the utility of network simulations without compromising privacy.

Findings

Both methods yield similar accuracy in bus-voltage magnitude estimation.

The load database allocation method effectively anonymises data while preserving network analysis accuracy.

Grouping and permutation method provides comparable results to existing approaches.

Abstract

Smart-meters are a key component of energy transition. The large amount of data collected in near real-time allows grid operators to observe and simulate network states. However, privacy-preserving rules forbid the use of such data for any applications other than network operation and billing. Smart-meter measurements must be anonymised to transmit these sensitive data to a third party to perform network simulation and analysis. This work proposes two methods for data anonymisation that enable the use of raw active power measurements for network simulation and analysis. The first is based on an allocation of an externally sourced load database. The second consists of grouping smart-meter data with similar electric characteristics, then performing a random permutation of the network load-bus assignment. A benchmark of these two methods highlights that both provide similar results in bus-voltage magnitude estimation concerning ground-truth voltage.