An Adaptive Bacterial Foraging Optimization Algorithm Based on Chaos-Enhanced Non-elite Reverse Learning

Bacterial Foraging Optimization Algorithm (BFO) is a swarm intelligence-based optimization algorithm that has been widely applied in various fields. However, the classical BFO still suffers from two major limitations: The fixed chemotactic step-size makes it difficult to balance exploration and exploitation capabilities, and the non-elitist elimination strategy in the reproduction phase may lead the population to be trapped in local optima. To address the two limitations of the classical BFO, this paper presents an improved BFO with adaptive chemotactic step-size and chaos-enhanced non-elite reverse learning (CLBFO). In CLBFO, the chemotactic step employs an adaptive nonlinear dynamic step-size strategy. Each bacterial individual adaptively selects an appropriate chemotactic step-size at different stages of the optimization process, which effectively alleviates the low search efficiency and oscillation problems caused by the fixed step-size. The reproduction step improves the non-elite solutions based on non-elite reverse learning and incorporates a chaotic disturbance mechanism to enhance the convergence speed and effectively reduce the possibility of the population falling into local optima. The performance of the proposed algorithm was evaluated on the benchmark test suite and compared with that of other intelligent optimization algorithms. The comparisons demonstrated the effectiveness of the proposed algorithm in balancing exploration and exploitation, and reducing the risk of local convergence.