Integrating opposition-based learning into the evolution equation of bare-bones particle swarm optimization

Abstract

Bare-bones particle swarm optimization (BPSO) is attractive since it is parameter free and easy to implement. However, it suffers from premature convergence because of quickly losing diversity, and the dimensionality of the solved problems has great impact on the solution accuracy. To overcome these drawbacks, this paper proposes an oppositionbased learning (OBL) modified strategy. First, to decrease the complexity of algorithm, OBL is not used for population initialization. Second, OBL is employed on the personal best positions (i.e., Pbest) to reconstruct Pbest, which is helpful to enhance convergence speed. Finally, we choose the global worst particle (Gworst) from Pbest, which simulates the human behavior and is called rebel learning item, and is integrated into the evolution equation of BPSO to help jump out local optima by changing the flying direction. The proposed modified BPSO is called BPSO-OBL, it has been evaluated on a set of well-known nonlinear benchmark functions in different dimensional search space, and compared with several variants of BPSO, PSOs and other evolutionary algorithms. Experimental results and statistic analysis confirm promising performance of BPSO-OBL on solution accuracy and convergence speed in solving majority nonlinear functions.

Simple PSO Algorithm with Opposition-based Learning Average Elite Strategy

“…In order to better evaluate the solving performance and efficacy of the proposed OLAE-SPSO algorithm, in this section,we select a suit of test functions such as Sphere, Schwefel's P2.22, Quadric, Griewank, Rastrigin and Ackley from reference [13]and [14],which can be expressed as 1 f , 2 f , 3 f , 4 f , 5 f , 6 f ,respectively.The first three functions ( 1 f -3 f ) are unimodal optimization functions which are used to investigate the convergence speed and optimization precision of the algorithms while the rest of the functions ( 4 f -6 f ) are multimodal optimization functions which are used to evaluate the abilities of jumping out of local optimum and seeking the global excellent result. The unimodal optimization functions only have one extreme value point within a given search area while the multimodal optimization functions have more than one local one extreme value points within a given search area.…”

“…Shi and Eberhart introduced particle swarm optimization algorithm with linear descend inertia weight, which uses the dynamic inertia weight that decreases linearly in the light of iterative generations increasing. In [5],the opposition-based learning(OBL) was employed on the personal best positions to reconstruct them, which is helpful to enhance convergence speed. Tsai et al combined the gravity search methods and PSO algorithm and gave a central role to the global optimal particle, which can improve the global convergence of the algorithm in [6].Zhou et alput forward to an improved particle swarm optimization combined with differential evolutionary mutation and elite opposition-based learning in [7].…”

Structural Damage Identification Based on l1Regularization and Bare Bones Particle Swarm Optimization with Double Jump Strategy

“…On the other hand, there are some researchers who focused on keeping swarm diversity and proposed some measures to enhance the diversity of the swarm, such as separate iteration [26] and dynamic local search strategies [29]. Liu et al [30] utilized a novel disruption strategy, originating from astrophysics, to shift the abilities between exploration and exploitation during the search process, and an opposition-based learning (OBL) modified strategy was further investigated [31].…”