Orthogonal learning particle swarm optimization

ABSTRACT

This paper proposes an orthogonal learning particle swarm optimization (OLPSO) by designing an orthogonal learning (OL) strategy through the orthogonal experimental design (OED) method. The OL strategy takes the dimensions of the problem as the orthogonal experimental factors. The levels of each dimension (factor) are the two choices of the personal best position and the neighborhood's best position. By orthogonally combining the two learning exemplars, the useful information can be discovered, preserved and utilized to construct an efficient exemplar to guide the particle to fly in a more promising direction towards the global optimum. The effectiveness and efficiency of the OL strategy is demonstrated on a set of benchmark functions by comparing the PSOs with and without OL strategy. The OL strategy improves the PSO algorithm in terms of higher quality solution and faster convergence speed.

REFERENCES

• 1: J. Kennedy and R. Eberhart, "Particle swarm optimization," in Proc. IEEE Int. Conf. Neural Networks, 1995, pp. 1942--1948.
• 2: Math. Stat. Res. Group, Chinese Acad. Sci., Orthogonal Design (in Chinese). Beijing: People Education Pub., 1975.

INDEX TERMS

Primary Classification:
  I. Computing Methodologies
  I.2 ARTIFICIAL INTELLIGENCE
      I.2.8 Problem Solving, Control Methods, and Search
          Subjects: Heuristic methods

General Terms:
Algorithms, Experimentation, Performance, Reliability

Keywords:
global numerical optimization, orthogonal experimental design, orthogonal learning particle swarm optimization, particle swarm optimization