Privacy-Preserving Aggregation of Virtual Power Plant Data via Differential Privacy and Secure Multiparty Computation

Virtual power plants (VPPs), a key paradigm in smart grids, depend on accurately aggregating the firm, deliverable capacity of distributed users, yet individual data are highly sensitive. To meet the need for privacy-preserving aggregation across users, this paper proposes a mechanism that integrates differential privacy (DP) with secure multiparty computation. Specifically, each user adds a fixed, high-magnitude noise term to its data and reuses it over the entire horizon to mask user’s characteristics; in addition, a small Gaussian perturbation is injected in every interval to provide DP for temporally adjacent records and bound inference risk. The fixed noise is secret-shared and exchanged on the user side, enabling the aggregator to efficiently reconstruct and subtract the aggregate fixed noise, thereby recovering a nearly unbiased total without access to any individual-level information. The mechanism offers strong privacy guarantees while maintaining high accuracy and efficiency, with controllable communication and computation overheads. It can be directly applied to support VPP market participation and operations-and-maintenance decision-making based on aggregated firm supply capability.