Adaptive Swarm Intelligence: A Time-Varying Fitness Guided Optimization Algorithm for Dynamic Search Strategies

Existing swarm intelligence algorithms have utilized fitness to select individuals and developed search strategies based on their locations.However, these algorithms overlook the dynamic variations in fitness over time, limiting the flexibility to adjust search step sizes and reducing optimization performance.To address this, a Time-varying Fitness Guided Optimization Algorithm (TFGOA) is proposed.TFGOA guides individuals to effectively adapt for global and local searches by quantifying timevarying fitness characteristics.Specifically, TFGOA proposes an individual time-varying fitness factor to detect real-time variations in performance.Based on this factor, a time-varying fitness guided search mechanism is developed.This mechanism flexibility adjusts the search behavior for each individual, thereby identifying potential optimal solutions more accurately.TFGOA is tested on the CEC2017, CEC2020, and CEC2022 benchmarks and is comprehensively compared with seven state-of-the-art algorithms.Experimental results demonstrate that TFGOA outperforms the comparison algorithms in terms of convergence and computational time.TFGOA is also tested on welded beam, gear train, and three-bar truss design problems, showing superiority over other comparison algorithms in terms of solution accuracy.