Federated learning-based tuning of smart inverter parameters for voltage control

The rapid growth of photovoltaic (PV) systems in distribution networks has increased the need for advanced voltage regulation strategies. Smart inverters (SIs), equipped with voltage control functions such as fixed power factor, Volt-Var, and Volt-Watt control, as well as communication capabilities with external systems, are expected to manage local voltage profiles by adjusting active and reactive power. A key challenge lies in determining which control logic and parameter settings are appropriate for each prosumer under varying grid conditions. To address this, we propose a data-driven optimisation framework based on personalised federated learning. Locally measured voltage and PV generation time series are compressed into statistical descriptors representing joint distributions, enabling evaluation of control strategies without detailed system models. The objective function balances total curtailment and fairness in PV output reduction, quantified through a generalised squared Hellinger distance. By coordinating parameter updates across prosumers with similar operating conditions, the approach achieves scalable and model-free optimisation. Simulation studies on a high-PV penetration network demonstrate that the method improves voltage regulation while ensuring fair utilisation of renewable generation capacity. Simulation studies on a high-PV penetration distribution network substantiate that the proposed method effectively mitigates PV output curtailment while ensuring voltage regulation capability.