摘要： 根据库恩的知识进化观，提出一种基于范式转换的知识进化算法。每个范式对应一个问题的可行解，以范式为单位建立初始知识库。利用传承算子实现对优秀范式的传承，采用修补算子实现范式危机的消除，以创新算子产生新范式，从知识库的最优范式中获取问题的最优解。将该算法应用于求解函数极小值，其结果与遗传算法相比具有更好的寻优性能。

关键词: 知识进化算法, 范式转换, 传承算子, 修补算子, 创新算子

Abstract: Based on Kuhn’s evolutionary epistemology idea, this paper proposes a knowledge evolutionary algorithm based on paradigm shift. The paradigm is according to the solution, and an initial knowledge base is formed. The next work is to inherit excellent knowledge individuals by inheritance operator, produce novel knowledge individuals by innovation operator, eliminate the crisis of paradigm by update operator, and accordingly realize knowledge evolution. The optimal solution of issues can be gained from the optimal knowledge individual. Experiments are taken on optimization of functions. Compared with genetic algorithm, the proposed algorithm can search the global optimal solution with less population and faster speed.

Key words: knowledge evolutionary algorithm, paradigm shift, inheritance operator, repair operator, innovation operator

中图分类号: 

• TP18