Machine learning assisted controller design for voltage regulation in a more electric aircraft power system

Abstract Three-stage synchronous generators (TSSG) are used in a more electric aircraft (MEA) to power various parts of the aircraft, such as environmental, hydraulic, avionics, and mechanical systems. However, regulating the voltage output of TSSGs in the presence of speed and load variations presents a significant challenge due to the dynamic couplings inherent in the system. In this work, a machine learning-assisted controller (MLAC) is designed to regulate the output voltage of the TSSG system at variable speeds. Moreover, data-driven techniques are employed for the training, testing, and deployment of the proposed MLAC controller. Furthermore, variants of meta-heuristics algorithms are investigated to fine-tune the response of the proposed controller through the selection of optimal hidden and output layer weights. Additionally, the transparency of the proposed controller is addressed and the optimized weights are auto-tuned with the assistance of a fuzzy logic controller (FLC). The resultant intelligent controller is evaluated in MATLAB/Simulink environment on a nonlinear model of the three-stage generator. The effectiveness and validity of the proposed approach in controlling the output voltage of the TSSG system are confirmed through comprehensive results analysis.