Decentralized Privacy-preserving Power Flow Forecast on Smart Meters

Severin Nowak (Lucerne Univeristy of Applied Sciences and Arts); Fabian Widmer (Lucerne Univeristy of Applied Sciences and Arts); Jeremy Taylor (Via Science); Antonios Papaemmanouil (Lucerne Univeristy of Applied Sciences and Arts); Daniel Raimundo (Lucerne Univeristy of Applied Sciences and Arts)

Abstract

The transition toward decarbonisation of energy systems necessitates significant adaptations in distribution networks, challenging the grid planning and operation for Distribution System Operators (DSOs). This paper explores how smart meters can improve DSO capabilities amidst evolving regulatory frameworks and technological landscapes. DSOs face the challenge of accommodating more distributed generation, electrification of loads, and the surge in electric vehicles, all while striving to maintain grid reliability and cost-effectiveness.
To meet these demands without compromising data security and privacy, advanced data science techniques are explored, leveraging smart meter data in near real-time.

The paper proposes decentralised algorithms capable of near real-time computation and forecasting of network power flows, revealing congestion issues without centralising consumer data. Field trials conducted in Rolle VD provide concrete evidence of the feasibility and advantages of this approach. They not only offer insights into network conditions but also ensure data privacy. Moreover, the developed decentralised congestion calculation method has shown to deliver results comparable to commonly used centralised algorithms, even when confronted with real-world challenges such as connection issues. Utilizing smart meters with enhanced computing capabilities and Internet of Things (IoT) infrastructure, the proposed method presents a viable alternative to traditional approaches, enabling DSOs to adapt to the evolving energy landscape efficiently.