APBAO: Adaptive and Parallel Beetle Antennae Optimization

Beetle Antennae Search Algorithm (BAS) is a new single-agent intelligent optimization algorithm, which low computational complexity, ease of implementation and fast convergence speed in low-dimensional function optimization. Based on BAS, this paper proposes Adaptive and Parallel Beetle Antennae Optimization Algorithm (APBAO), which evolves from a single iterative individual in BAS to multiple parallel iterative individuals, improving algorithm efficiency through adaptive step size. Finally, putting forward the elite system, which preserves high-quality solutions in each iteration. To verify the performance of the algorithm, this paper conducts tests using multiple standard benchmark functions and compares APBAO with BAS, Particle Swarm Optimization Algorithm (PSO) and Ant Colony Optimization Algorithm (ACO). The experimental results show that APBAO improves performance by 97.39% compared to BAS, and by 84.46% and 86.98% compared to the PSO and ACO, respectively.

I. Introduction

Optimization has been always concerned and enumeration methods have been proposed to solve different kinds of optimization problems[1]. One of the most significant current discussions in optimization is how to obtain optimal solutions[2] for complex functions or functions with various constraints, wherein corresponding to actual problems in reality[3]–[5]. Therefore, many stochastic metaheuristics has been proposed with the principles of natural selection to guide a set of solutions toward the optimal one[6], [7]. Among them, Beetle Antennae Search Algorithm (BAS) is a novel bio-inspired heuristic optimization algorithm with the idea derived from the observation and study of the foraging process of beetles[8]. BAS has the advantages of few parameters, low computational complexity and easy implementation. BAS converges quickly in low-dimensional optimization functions and can be used to solve complex nonlinear optimization problems[9]. However, basic BAS has a lot to improve on[10], [11], to make it more effective and efficient in searching optimal solution to practice.