Optimal Photovoltaic Generation Allocation for Minimum Active Power Loss in Distribution Systems Based on an Improved Backtracking Search Optimization Algorithm

PVG allocation generally involves optimal PV power sizing and locations for PVG installations in power distribution networks, which can be formulated as single or multiple related objective functions under specific power system constraints. This paper proposes an improved backtracking search algorithm based on two new operators. It compares the ability of four different metaheuristic algorithms, namely, a genetic algorithm, backtracking search optimization, data-mining backtracking search optimization algorithm developed with mutual learning, and bird swarm algorithm, to optimize the placement and sizing of PVGs in IEEE 33-bus and 69-bus power distribution systems. Based on BSOA, DBSOAdML is developed by enhancing the search capability with two new operators that mutually investigate the fitness and population in the adjacent generation. In this study, the location and power sizing of PVG in the distribution system are optimized for the minimum power loss in terms of multiple PVG power sizings and their locations. The convergent performances of different algorithms are summarized and compared for four cases with increasing PVG installation. The results show DBSOAdML provides effective convergence outcomes and has a relatively faster speed over the other algorithms in terms of optimizing multiple PVG units for loss minimization in the 33-bus and the 69-bus systems