Self-Protective Inverters for Enhancing the Security of Smart Power Grids

This paper presents a self-protective algorithm employing analytical steady-state and hybrid dynamic models to detect malicious setpoints for grid-interactive inverters. The hybrid dynamic model is achieved by integrating an analytical risk factor into a neural network framework. The proposed self-protection strategy evaluates incoming setpoints from utility operators or third-party aggregators using reference models before applying them to the local controller. In cases where malicious setpoints bypass existing security layers, the smart inverter verifies the integrity of incoming setpoints using the reference models. The effectiveness of the self-protection strategy is assessed through experimental validation on a laboratory setup comprising a three-phase 3 kVA inverter and a 12-kW regenerative grid emulator. Results demonstrate that the hybrid dynamic models enhance accuracy and offer device-level protection for grid-interactive inverters, effectively preventing harmful setpoints from engaging with the local controller.