Modern electrical distribution networks face mounting pressure to reduce technical losses and maintain voltage stability while integrating increasing levels of distributed generation. Traditional optimization approaches often treat medium-voltage and low-voltage networks as isolated systems, missing opportunities for coordinated improvements and mutual voltage support between levels.
Researchers have proposed a novel multi-agent system architecture that simultaneously optimizes both network levels while determining optimal distributed generation capacity. The key innovation lies in recognizing that reconfiguration decisions at the medium-voltage level directly influence low-voltage network performance, and vice versa. By modeling these interdependencies explicitly, the system can exploit voltage support opportunities across the entire distribution network.
The methodology incorporates realistic operating conditions by accounting for uncertainties in renewable generation output and customer demand variability. Network reconfiguration is achieved through optimal switch operations—a practical mechanism already present in modern distribution networks—combined with strategic placement and sizing of distributed resources.
Validation on a test system comprising one medium-voltage network feeding three low-voltage feeders demonstrates substantial improvements. Medium-voltage losses decreased 44.7% compared to the initial configuration and 25.8% relative to reconfigured networks without distributed generation. The integrated medium-voltage and low-voltage approach further reduced total system losses by 5.57% compared to centralized optimization methods.
Beyond loss reduction, the system maintains nearly uniform voltage profiles throughout the network during varying operating conditions. This stability is crucial for protecting sensitive equipment and ensuring power quality for connected customers.
The multi-agent architecture offers practical advantages over centralized approaches, enabling distributed decision-making that aligns with modern, increasingly autonomous distribution network operations. Implementation using standard tools like MATLAB and MATPOWER ensures accessibility for utilities considering deployment. As distribution networks continue evolving toward smarter, more flexible configurations, such integrated optimization methods will become essential for maximizing efficiency while maintaining stability and reliability.
