Maintaining Genetic Diversity in Bacterial Evolutionary Algorithm

The Bacterial Evolutionary Algorithm (BEA) is a relatively new type of evolutionary algorithm and shows the typical phenomena of stochastic optimization methods. Two of these phenomena: premature convergence and low convergence speed near the optimum are often in connection with the low genetic diversity of the population. Variation of genetic diversity in the original BEA and in its three parallel variants have been examined and documented. Several possible ways of increasing the diversity have been also studied. In this paper the authors present an effective method called forced mutation to maintain the genetic diversity of the population and to speed up the convergence. With forced mutation a significant improvement has been achieved in all test cases, but the parameter setting is problem-dependent. Therefore a possible way of adaptive parameter setting for forced mutation is also proposed.