<![CDATA[Pico hydropower is a renewable-energy option for isolated communities and low-head run-of-river sites, but axial-flow pico-hydro generators are vulnerable to voltage fluctuation when water flow, hydraulic head, or consumer load changes. This study proposes a novel and reproducible artificial-intelligence-assisted proportional-integral-derivative (PID) tuning framework for voltage control of a 220 V, 2 kW axial-flow turbine generator ZD760-LM-(18-20). The novelty lies in combining a voltage-control-oriented small-signal model of a low-head axial-flow pico-hydro unit, a nonminimum-phase hydraulic zero that represents inverse initial response, identical bounded PID-search constraints, and a composite objective that explicitly penalizes inverse dip, overshoot, settling time, ITAE, and IAE. The plant model combines actuator or electronic-load-controller dynamics, non-elastic water-column dynamics, turbine-generator dynamics, and sensor dynamics. PID gains obtained from Ziegler-Nichols (PID-ZN), Ant Colony Optimization (PID-ACO), and Particle Swarm Optimization (PID-PSO) are compared under Kp = 0-100, Ki = 0-50, and Kd = 0-10. Simulation results show that PID-ZN stabilizes the plant but requires a 6.80 s settling time and produces an ITAE of 2.9603. PID-ACO reduces settling time to 2.26 s and ITAE to 1.1320, whereas PID-PSO gives the lowest ITAE of 1.1311 with only 0.030% overshoot. Compared with PID-ZN, PID-PSO reduces settling time by 66.8% and ITAE by 61.8%. These results indicate that AI-based PID tuning can improve voltage quality in low-cost rural and off-grid pico-hydro systems using practical ELC or simple actuator implementations.]]>
Copyrights © 2026
