An Adaptive Differential Evolution with Unsymmetrical Mutation

Differential Evolution (DE) is one of the evolutionary algorithms under active research. It has been successfully applied to many real world problems. In this paper, an improved DE with a novel mutation scheme is proposed. The improved DE assigns a distinct scale factor for each individual mutation based on the fitness associated with each base vector involved in the mutation. With the adoption of different scale factors for mutation, DE is capable of searching more locally around superior points and explore more broadly around inferior points. Consequently, a good balance between exploration and exploitation can be achieved. Also, an adaptive base vector selection scheme is introduced to DE. This scheme is capable of estimating the complexity of objective functions based on the population variance. When the problem is simple, it will tend to select good vectors as base vector which will lead to quick convergence. When the objective function is complex, it will select base vector randomly so that the population maintains a high exploration capability and will not be trapped into local minima so easily. A suite of 12 benchmark functions are used to evaluate the performance of the proposed method. The simulation result shows that the proposed method is promising in terms of convergence speed, solution quality and stability.