Multistage Ant System Optimization Algorithm

Abstract

In this paper we introduce a modified ant system optimization algorithm, we call it the multistage ant system optimization (MS-ASO). This modified algorithm have many advantageous like releasing ants from local optima, enabling ants to accept temporary bad solution in order to find better one later, improving the diversification by forcing (not encouraging) ant to search new parts of the solution space, and also pass information about the global best solution between the predetermined number of iterations/stages. We test our modified algorithm on some benchmark problems of the traveling sales man problem to see how good it is, the result we get was encouraging that is we succeed in reducing the number of calculations (iterations) and we also find new optimum solution (i.e. routes) for some benchmark problems which are better (i.e. shortest) than published routes.

Keywords

• traveling seals man problem
• ant system optimization