Data Distribusi Beban

Load imbalance in distribution transformers is a critical issue that could affects the efficiency, power losses, and reliability of low-voltage distribution systems. In Power Company X, initial observations suggest that 59 transformers appear to operate in overload conditions, while others remain significantly underutilized. This imbalance suggests that a more adaptive and data-driven optimization approach. This study proposes a hybrid optimization model by integrating the K-Nearest Neighbor (KNN) algorithm, Monte Carlo simulation, and the Non-dominated Sorting Genetic Algorithm II (NSGA-II). KNN is used to identify candidate transformers based on spatial proximity and customer load characteristics. Monte Carlo simulation is used to generates diverse initial populations to support the evolutionary search process. Subsequently, NSGA-II is applied to optimizes two main objective functions: minimizing load variance across transformers and minimizing the average customer-to-transformer distance. The number of overloaded transformers is reduced by 80%, and the average distance between customers and assigned transformers decreases by 23%. Furthermore, the model appears to exhibits strong robustness against demand uncertainty, as shown by the sensitivity analysis conducted under load increases up to 7%, with performance degradation remaining minimal. These findings indicate that the proposed hybrid approach could effectively enhance load balancing and spatial efficiency. The model could serve as a promising foundation for developing an artificial intelligence based decision support system for distribution planning within Company.