Reactive Power Optimization Based on Improved Bird Swarm Algorithm

This paper proposes a modified bird swarm algorithm, which introduces levy flight strategy and sorting into bird swarm algorithm (LSBSA). The levy flight strategy ensures initial swarm diversity, producers and beggars are classified by sorting method rather than random strategies, therefore accelerates convergence rate of BSA algorithm. Four benchmark functions involving unimodal and multimodal search problems are used to test the performance of the proposed BSA. Simulation results are compared with these of BSA, inertia weight particle swarm optimization (IWPSO), and artificial bee colony algorithm (ABC). It is indicated that proposed LSBSA is better than other three algorithms, in terms of convergence rate and accuracy of solution. Moreover, the proposed algorithm is applied to reactive power optimization in IEEE 3D-bus system. The goal is to minimize active power loss. Parameters of generator, transformer, and compensator are selected as control variables. Optimal solution testifies the feasibility and availability in such practical engineering.