Privacy-preserving utilization of distribution system flexibility for enhanced TSO-DSO interoperability: A novel machine learning-based optimal power flow approach

Power system transformation makes distribution system (DS) flexibility crucial for efficient network management. Leveraging this flexibility requires interoperability between Transmission System Operators (TSOs) and Distribution System Operators (DSOs). However, data privacy concerns pose significant challenges to the effective utilization of this flexibility, since its integration often requires the exchange of sensitive information between TSOs and DSOs. For instance, in a conventional AC optimal power flow (OPF) problem, the TSO requires access to sensitive DSO information, such as network topology. To address this, we propose a machine learning (ML) based method in which DSOs train ML models using datasets that do not contain sensitive data, resulting in models defined by non-sensitive parameters. This prevents the transfer of sensitive information. Because models are trained solely on non-sensitive data, sensitive information remains protected against reverse engineering. After these trained models are shared by the DSOs with the TSO, the TSO can solve the OPF problem and determine flexibility-providing unit (FPU) dispatch in a single communication round. To achieve this, we introduce a tailored neural network (NN) architecture to efficiently represent the DS feasible region. To assess the effectiveness of the proposed method, we benchmark it against the standard AC-OPF on multiple DSs with meshed connections and multiple points of common coupling (PCCs) with varying voltage magnitudes. The numerical results show that the proposed method achieves competitive performance while preserving data privacy. Additionally, since this method directly determines the dispatch of FPUs, it eliminates the need for an additional disaggregation step. Overall, the proposed approach enables the effective utilization of DS flexibility in network management without compromising data privacy, thereby enhancing interoperability among stakeholders.