Model-Free Privacy Preserving Power Flow Analysis in Distribution Networks

TL;DR

This work tackles privacy concerns in model-free distribution-network power flow by introducing a privacy-preserving framework that combines Local Randomization of SM data, zero-knowledge proof–based data collection for secure voltage datasets, and incremental learning to adapt to seasonal load variations. The approach enables training of ANN-based PF models without exposing household data, while maintaining high accuracy even under measurement errors. Key contributions include the LRS for irreversibile, per-user data obfuscation, a ZKP-based protocol for secure voltage data collection, and an IL strategy triggered by a Wasserstein-distance indicator to sustain performance over time. The framework demonstrates practical viability with fast data collection (one month in under an hour), robust voltage estimation (mean errors around 0.005–0.014 p.u.), and adaptability across multiple low-voltage networks and seasonal patterns, suggesting a scalable path to privacy-preserving model-free PF deployment.

Abstract

Model-free power flow calculation, driven by the rise of smart meter (SM) data and the lack of network topology, often relies on artificial intelligence neural networks (ANNs). However, training ANNs require vast amounts of SM data, posing privacy risks for households in distribution networks. To ensure customers' privacy during the SM data gathering and online sharing, we introduce a privacy preserving PF calculation framework, composed of two local strategies: a local randomisation strategy (LRS) and a local zero-knowledge proof (ZKP)-based data collection strategy. First, the LRS is used to achieve irreversible transformation and robust privacy protection for active and reactive power data, thereby ensuring that personal data remains confidential. Subsequently, the ZKP-based data collecting strategy is adopted to securely gather the training dataset for the ANN, enabling SMs to interact with the distribution system operator without revealing the actual voltage magnitude. Moreover, to mitigate the accuracy loss induced by the seasonal variations in load profiles, an incremental learning strategy is incorporated into the online application. The results across three datasets with varying measurement errors demonstrate that the proposed framework efficiently collects one month of SM data within one hour. Furthermore, it robustly maintains mean errors of 0.005 p.u. and 0.014 p.u. under multiple measurement errors and seasonal variations in load profiles, respectively.

Figures (10)

• Figure 1: Framework of ANN-based PF calculation with local privacy-preserving strategy: orange, blue, and red lines represent load transfer, voltage transfer, and the communication channel, respectively.
• Figure 2: Nonlinear transformation function without constraints in (a) and with constraints in (b).
• Figure 3: Framework of the ZKP-based data collection process: red lines indicate communication paths and blue lines illustrate the process steps.
• Figure 4: Illustrative example of uniformly distributed vectors construction
• Figure 5: Distribution of original $P$ and transformed dataset $P^*$ (in the first and second row) and their correlations (in the third row).