Abstract
This article addresses the problem of sensorless control of a squirrel-cage induction generator (SCIG) used as
a shaft generator in shipboard DC power systems. The work is motivated by the growing adoption of DC ship-
board microgrids, where robust and cost-effective generators without mechanical sensors are desirable. A math-
ematical model of the SCIG in α-β and d-q reference frames is derived and discretized, enabling efficient re-
al-time implementation on a DSP–FPGA control platform. Building on this model, the paper proposes a hybrid
current–voltage observer that combines a current-model estimator (effective at low speed) with a voltage-model
estimator (accurate at medium and high speed). PI-based compensating voltages are introduced to suppress
numerical integration drift and reduce sensitivity to stator resistance variation, ensuring accurate estimation
of rotor flux, flux angle, and rotational speed. The excitation process of the generator is analysed with respect to
initial DC-link voltage, shaft speed, and loss currents, and criteria for successful self-excitation are identified.
The proposed algorithm is validated through time-domain simulations and laboratory experiments on a 3 kW
SCIG system. Results show close agreement between simulation and measurements, stable DC-link voltage
regulation under varying load and speed, and reliable sensorless operation that is comparable to, or more robust
than, encoder-based systems in the presence of EMI. The study confirms the practical suitability of the hybrid
observer for shipboard SCIG applications over a wide rotational speed operating range.