Neural Computing and Applications
This paper presents an improved coyote optimization algorithm (ICOA) for the electric distribution network reconfiguration (EDNR) problem considering unbalanced load. ICOA is first developed by carrying out two improvements on two new solution generation mechanisms of original coyote optimization algorithm (COA). In the first mechanism, ICOA has used the so-far best solution instead of the tendency solution like COA. In the second mechanism, a local search mechanism has been proposed to update the so-far best solution. ICOA determines opened switches in aim to minimize total power losses. In addition, a modified power flow (MPF) method based on the technique of backward/forward sweeps is proposed to solve power flow for unbalanced distribution system. The proposed MPF method has been highly accurate in comparison with the Power System Simulator/Advanced Distribution Engineering Productivity Tool software (PSS/ADEPT). The ICOA together with COA, particle swarm optimization (PSO) and sunflower optimization (SFO) have been implemented on three systems including 25-node, 33-node and 69-node for comparison. As a result, ICOA has outperformed COA, PSO, SFO and other existing methods for the EDNR problem. Consequently, the combination of the proposed MPF method and ICOA for solving EDNR problem with unbalanced load can lead to high effectiveness.
