Enhanced Bacterial Foraging Optimization with Dynamic Disturbance Learning and Bilayer Nested Structure

Traditional bacterial foraging algorithm often struggle with the limitations of prematurely converging to local optimums, sluggish execution processes and higher memory consumption. To address these problems, a bilayer nested bacterial foraging algorithm incorporating a dynamic disturbance learning strategy (BiddBFO) is proposed. The novel algorithm incorporates three update strategies: i) The piecewise linear chaotic map (PWLCM) was used to initialize the bacterial population, ensuring robust spatial traversal capabilities. ii) A dynamic reverse learning strategy was introduced in the chemotaxis operation to facilitate variant fusion, increasing the chance of information exchange between bacteria and prevent falling into local optima. iii) A bilayer nested structure was designed, which not only reduces the number of nesting layers and the overall runtime of the BFO but also preserves the inherent characteristics and benefits of the original algorithm. Finally, we conducted extensive testing on six benchmark functions and compared its performance against five other algorithms. The results demonstrated that the BiddBFO algorithm achieved superior convergence accuracy and a faster convergence rate for the objective function.