A Lyapunov-Stable Direct Deadbeat Control Strategy for Grid-Current-Sensor-Only Active Power Filters

To improve the reliability and precision of shunt active power filters (APFs) under disturbances, this paper proposes an enhanced direct deadbeat control strategy requiring only grid-side current sensors. To this end, a sensor-lean yet robust framework is established by integrating PLL-less voltage estimation with online inductance identification. Specifically, the need for AC voltage sensors is eliminated by reconstructing the grid voltage from inverter outputs and consecutive current samples, while a load current feedforward mechanism further obviates the load current sensors. From an algorithmic perspective, the strategy utilizes the grid-side current as the direct controlled variable to minimize error propagation, while an online identification algorithm is incorporated to counteract parameter drift induced by magnetic saturation. Furthermore, system stability is rigorously guaranteed via Lyapunov theory. Validation through both simulation and experiments reveals that the grid current THD is reduced to 2.90% and 3.3%, respectively, with a dynamic response time within 20 ms. Ultimately, these findings confirm that the proposed scheme minimizes hardware dependency without compromising harmonic suppression or transient robustness.