An Effective Bacterial Foraging Optimization Based on Conjugation and Novel Step-Size Strategies

Abstract

Bacterial Foraging Optimization (BFO) is a high-efficient meta-heuristic algorithm that has been widely applied to the real world. Despite outstanding computing ability, BFO algorithms can barely avoid premature convergence in computing difficult problems, which usually leads to inaccurate solutions. To improve the computing efficiency of BFO algorithms, the paper presents an improved BFO algorithm: Conjugated Novel Step-size BFO algorithm (CNS-BFO). It employs a novel step-size evolution strategy to address limitations brought by fixed step size in many BFOs. Also, the improved BFO algorithm adopts Lévy flight strategy proposed in LPBFO and the conjugation strategy proposed in BFO-CC to enhance its computing ability. Furthermore, Experiment on 24 benchmark functions are conducted to demonstrate the efficiency of the proposed CNS-BFO algorithm. The experiment results suggest that the proposed algorithm can deliver results with better quality and smaller volatility than other meta-heuristics, and hence sufficiently mitigate the limitation of premature convergence facing many meta-heuristics.