Differential Duffing Oscillator-Based Cyberattack Detection for Inverters

A differential Duffing oscillator detector (DDOD) based method is developed to detect controller manipulation and replay attacks on inverters without compromising networked microgrids' operations. Specified probing signals are injected into the inverter controllers and the responses are then detected by the DDOD to identify attacks through the change of its operating states. Our contributions include: 1) A lightweight model-free DDOD for real-time and accurate localization of attacks; 2) A weighted moving average-based method to integrate controller parameters as coupling coefficients of the DDOD for attack type identification. Experimental results validate the efficacy of the devised detection method and verify DDOD's robustness under low signal-to-noise ratio conditions.