A Fast-Convergent Hyperbolic Tangent PSO Algorithm for UAVs Path Planning

Abstract

Particle Swarm Optimization (PSO) stands as a cornerstone among population-based swarm intelligence algorithms, serving as a versatile tool to tackle diverse scientific and engineering optimization challenges due to its straightforward implementation and promising optimization capabilities. Nonetheless, PSO has its limitations, notably its propensity for slow convergence. Traditionally, PSO operates by guiding swarms through positions determined by their initial velocities and acceleration components, encompassing cognitive and social information. In pursuit of expedited convergence, we introduce a novel approach: the Cognitive and Social Information-Based Hyperbolic Tangent Particle Swarm Optimization (HT-PSO) algorithm. This innovation draws inspiration from the activation functions employed in neural networks, with the singular aim of accelerating convergence. To combat the issue of slow convergence, we reengineer the cognitive and social acceleration coefficients of the PSO algorithm, leveraging the power of the hyperbolic tangent function. This strategic adjustment fosters a dynamic balance between exploration and exploitation, unleashing PSO's full potential. Our experimental trials encompass thirteen benchmark functions spanning unimodal and multimodal landscapes. Besides that, the proposed algorithm is also applied to different UAV path planning scenarios, underscoring its real-world relevance. The outcomes underscore the prowess of HT-PSO, showcasing significantly better convergence rates compared to the state-of-the-art.

Keywords

• Particle swarm optimization (PSO)
• acceleration coefficients
• convergence rate
• exploration
• exploitation
• and path planning