An Adaptive Artificial Bee Colony Algorithm with Success-Driven Strategy Selection and Dimension Modification

The Artificial Bee Colony (ABC) algorithm demonstrates strong exploration but converges slowly. To address this, AMRABC_Ts introduces an adaptive dimension modification rate mechanism with two search strategies. The exploration-favoured CABC strategy uses random guidance, while the exploitation-favoured MABC_order strategy employs ranking information from elite solutions. These strategies are dynamically selected based on improvement feedback to maintain search balance throughout evolution. Dimension modification rates are chosen according to historical success rates to accelerate convergence. Tests on 22 benchmark functions and CEC2014 problems show that AMRABC_Ts performs competitively against existing ABC variants at different dimensions, with notable strength on composition functions. The results validate improvements in convergence speed and solution accuracy, demonstrating the effectiveness of combining adaptive strategy selection with dimension modification.