A Novel Evolutionary Algorithm for Numeric Optimization

Since evolutionary algorithms (EA) were proposed as an optimization method, they have been used to solve problems in many fields. However, accuracy, convergence and stability are always topics in EA research. This paper shows our efforts in such direction. We employ a novel approach, namely the cloud model, to model evolution process and then present an efficient EA, i.e. the cloud model based evolutionary algorithm or CEA. Because the cloud model is very successful in uncertainty modeling and transition between qualitative concept and quantitative description, using it we can model inheritance and mutation during individuals´ evolution in a uniform and natural way. This enables CEA to perform well in fast converging to optimal values. Experiment results verify that CEA is much better than the compared algorithms, in terms of both precision and consistency.