A New Hidden Markov Model Approach for Pheromone Level Exponent Adaptation in Ant Colony System

We propose in this paper a Hidden Markov Model (HMM) approach to avoid premature convergence of ants in the Ant Colony System (ACS) algorithm. Indeed, the proposed approach was modelled as a classifier method to control the convergence through the dynamic adaptation of the α parameter that weighs the relative influence of the pheromone. The implementation was tested on several Travelling Salesman Problem (TSP) instances with different number of cities. The proposed approach was compared with the standard ACS and the existing fuzzy logic in the literature. The experimental results illustrate that the proposed method shows better performance. © 2020, The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG.