This paper investigates the effectiveness of a proposed approach for local power control of distributed generation (DG) units used for power and voltage support in passive distribution networks under high uncertainty in load and unexpected daily load variations. The approach is based on the use of an intelligent algorithm, Particle Swarm Optimization (PSO), to determine an asymptotic reference for the optimal real-time power control of small-scale DGs. The optimization problem is formulated with the desired objective functions, subject to network constraints. The principle of the proposed strategy is to determine the reference voltage at the connection nodes of the distributed generators (DGs) corresponding to optimal power generation, either as a fixed value or by deriving a regression line from a two-variable dataset comprising the DG’s optimal power generation and the voltage at its connection node. The optimal regression line is derived using the statistical least squares method. The optimal power–voltage datasets are obtained from various random load scenarios by applying the Particle Swarm Optimization (PSO) algorithm. In this study, the optimization algorithm is applied to minimize power losses and improve the voltage profile. The investigation includes simulation results and analysis conducted on a modified IEEE 33-bus radial distribution system with DG units providing both active and reactive power. The impact of active and reactive power generation on the objective functions is analyzed. Asymptotic reference values for the local control of DG units are derived for different scenarios, and the results demonstrate the effectiveness of the proposed strategy.