Abstract:

The particle swarm optimization (PSO) algorithm which is easy to converge to local optimum has a low convergence speed, a low precision and a big computation amount. Firstly, a new normalized PSO (NPSO) algorithm which does not need to detect whether the particles cross the border is proposed. The NPSO algorithm has a better validity and stability, a faster optimization speed, a better convergence precision and less computation than the PSO algorithm. Then, by combining wandering behavior of the wolf pack algorithm (WPA), a new DPSO (WPA-DPSO) algorithm which makes each partite particle has different exploration directions is proposed. The WPA-DPSO algorithm has the three-layer optimization ability, not only expanding the search range of the particle, and avoiding the algorithm converges to local optimum, but also effectively improving convergence speed, precision,validity and stability of the DPSO algorithm. Based on NPSO and the WPA-DPSO algorithm, a mixed PSO (WPA-NDPSO) algorithm is proposed. The WPA-NDPSO algorithm effectively avoids these problems of the PSO algorithm: earliness to convergence, small search range, easy to converge to local optimum, big computation amount. The simulation results of uniform linear array show that the WPA-NDPSO algorithm not only has a better convergence speed and precision, but also an excellent validity and stability.