Integrated learning particle swarm optimizer

Abstract

Particle Swarm Optimization (PSO) is population based stochastic optimization algorithm used in wide range of applications. PSO shares many similarities with Evolutionary Computation Techniques. PSO variants still finds difficulty to optimize the non-differentiable, nonlinear and complex multi-modal objective functions. This paper presents an improved PSO variant known as the Integrated Learning Particle Swarm Optimization (ILPSO) algorithm. The ILPSO approach captures the particles which are trapped in deep local minima, responsible for premature convergence and directs them towards optimal solution. This novel technique also introduces the updating strategy based on Hyperspherical coordinates of the particles to tackle with evenly distributed multiple minima of objective function. The proposed technique integrates Comprehensive Learning strategy to explore the solution effectively. The simulation comparisons of search efficiency, on the proposed ILPSO algorithm, are provided against different high quality PSO variants on the set of standard benchmark functions. The simulation results indicate that the proposed algorithm has high quality solution and faster convergence for complex and higher dimensional objective functions over traditional PSO. Copyright © 2007 IICAI.