A new charged ant colony algorithm for continuous dynamic optimization

Real world problems are often of dynamic nature. They form a class of difficult problems that metaheuristics try to solve at best. The goal is no longer to find an optimum for a defined objective function, but to track it in the search space. In this article we introduce a new ant colony algorithm aimed at continuous and dynamic problems. To deal with the changes in the dynamic problems, the diversification in the ant population is maintained by attributing to every ant a repulsive electrostatic charge, that allows to keep ants at some distance from each other. The algorithm is based on a continuous ant colony algorithm that uses a weighted continuous Gaussian distribution, instead of the discrete distribution, used to solve discrete problems. Experimental results and comparisons with two competing methods available in the literature show best performances of our new algorithm called CANDO on a set of multimodal dynamic continuous test functions.