Investigation of particles behaviors of piecewise-linear particle swarm optimizer
Tomoyuki Sasaki, Hidehiro Nakano · 2017
In our previous study, we have proposed piecewise-linear particle swarm optimizer (PPSO). The dynamics of each particle in PPSO has two search modes that are dynamically switchable to each other. PPSO has better search performance to solve non-separable problems than the classical PSO. However, since it was not clarified why PPSO was effective in solving such problems, we have now studied whether particle behavior in PPSO affects the search performance, focusing on search direction of particles to compare the search direction of PPSO particles with that of PSO particles. We further compared the search performance for parameter pattern of PPSO with the classical PSO based on numerical data. Here, we suggest that PPSO particles can move in solution space freely and that the behavior of PPSO particles is effective for solving the non-separable problems.