Aiming at addressing the issues of energy efficiency loss and quality fluctuation caused by insufficient electrode current control accuracy in the production process of electricity melt magnesium, a variable domain fuzzy PID control strategy based on multi-population genetic algorithm (MPGA) optimization is proposed. Firstly, the adaptive adjustment of the fuzzy domain is realized by constructing the dynamic contraction–expansion factor function, and the control accuracy under small deviation conditions is enhanced. Then, a multi-population co-evolution and elite migration mechanism is designed to optimize the controller parameters, which overcomes the local premature convergence issues of the traditional single-population genetic algorithm. Finally, comparative experiments with conventional control methods demonstrate that the proposed strategy significantly improves key performance metrics, including adjustment accuracy, response speed and anti-interference ability. Thus, the robustness of the system is significantly improved under complex conditions of raw material composition fluctuation and load disturbance. This methodology offers a novel approach for precise control and energy efficiency optimization of electricity melt magnesium smelting process.