Comparative assessment of classical, metaheuristic, and AI-based controller design and tuning strategies for VSC applications

Stable DC-link voltage regulation is essential for reliable power transfer and secure operation of Voltage Source Converter (VSC)-based systems. This paper presents a comprehensive comparative assessment of classical, metaheuristic, and AI-based controller design and tuning strategies for DC-link voltage regulation in a Voltage Source Converter. A single VSC terminal of a multi-terminal DC (VSC–MTDC) system is modeled and simulated in MATLAB/Simulink to assess the performance of different controller design and tuning strategies. The evaluated approaches include Symmetrical Optimum (SO), Pole Placement (PP), Ant Colony Optimization (ACO), Particle Swarm Optimization (PSO), Long Short-Term Memory (LSTM), and Fuzzy Logic Controller (FLC). Controller performance is evaluated using rise time, peak time, settling time, and percentage overshoot. The resulting controllers are subsequently evaluated in a four-terminal VSC–MTDC network using PSCAD/EMTDC under different operating conditions, including load changes, power-reference variations, reactive-power variations, operating-mode transitions, and regulating-terminal outage, to assess their dynamic performance under the specified operating conditions. The results show that the classical and metaheuristic approaches generally provide faster step responses, with ACO and PSO exhibiting the shortest settling times, while the AI-based LSTM and FLC controllers exhibit lower overshoot values but with comparatively longer settling times. The comparative results demonstrate distinct trade-offs between response speed and voltage-deviation characteristics among the evaluated control strategies.