A Bottleneck Assigned Binary Ant System for multimodal optimization

A bottleneck assigned binary ant system (baBAS) is presented and applied to multimodal optimization problems. The strategy is inspired by traffic organization in real ants under crowded conditions found by A. Dussutour et. al. The Strategy can prevent loss of diversity of solutions by introducing a so called ¿bottleneck¿ to each path of solutions. Any ant can´t pass through the path if the traffic flow on it has reached or exceeded the threshold. Experiment results on some typical multimodal complex functions showed that baBAS had much better capability than binary ant system (BAS), binary ant system with elitist strategy (BASe) and some traditional multimodal methods (Niche GA) in locating and marinating stable global and local optima in multimodal search space.