Flow Direction Algorithm with Lévy Distribution and Random Opposition for Simultaneous Network Reconfiguration with DGs and SCs Integration

Abstract

This article proposes an enhanced Flow Direction Algorithm (FDA) that integrates Lévy distribution and Random Opposition-Based Learning (ROBL) strategies to improve overall performance. First, the Lévy distribution is used to update the flow velocity vector, enabling a more accurate representation of movement toward the neighbour with the smallest objective function value, while introducing greater flexibility and randomness into velocity prediction. Second, ROBL refines the decision-making process during flow direction identification, enhancing the algorithm's adaptability. This promotes more effective exploration of potential solutions and strengthens convergence toward the optimal flow direction. The proposed ROFDA is comprehensively validated on simultaneous Optimal Network Reconfiguration (ONR) problems involving the integration of Distributed Generators (DGs) and Shunt Capacitors (SCs). Results clearly demonstrate that jointly optimizing network reconfiguration with DGs and SCs integration achieves significantly greater power loss minimization than optimizing each approach individually. Evaluated on standard 33-bus and 69-bus test systems, ROFDA consistently outperforms existing variants, including FDA, OFDA, and MFDA, thereby confirming its effectiveness, robustness, and superiority as an advanced optimization framework for solving complex power distribution network problems.