Mixed application of two learning mechanisms in genetic algorithm

For accelerating the algorithm convergence and avoiding the local optimization, an individual learning mechanism is often applied to generic algorithm to improve algorithm performance. The usual individual learning mechanism includes two sorts: Lamarckian learning and Baldwinina learning. The advantages and disadvantages of both mechanisms are indicated according to their difference performance in the generic algorithm. Additionally, based on a novel concept, named learning potentiality, and the method of digging individual learning potentiality by Baldwinina learning, the Lamarckian learning and Baldwinina learning are appropriately integrated for better algorithm performance so that the advantages of learning could be sufficiently utilized and disadvantages could be effectively forbidden. Numerical experimental results indicate the excellent effectivity of the integrated algorithm.