摘要
计算智能算法的融合可以有效解决实际问题,但算法选择带有一定盲目性。文章对计算智能的主要算法——人工神经网络、人工免疫系统、模糊系统和遗传算法等的特性进行比较,提出了四种融合形态——串联型、并联型、部分融合型和完全融合型,以及融合步骤、融合的数学描述,讨论了六种融合算法的特点和方法。融合提高了算法性能、扩大了应用范围。通过比较明确了计算智能算法的选择方法和进一步研究的方向;通过仿真分析说明了算法融合思路的正确性。
The fusion of computational intelligence＇s algorithms may be able to solve actual problems, but the method of selecting the algorithms may not be so scientific. The characteristics of four major algorithms-artificial neural network, artificial immune system, fuzzy logic system, and genetic algorithm-are compared in this paper. Fusion steps, fusion algorithm definition, and four kinds of fusion shapes （ seties, parallel, partial, and complete） are proposed. The characteristics and methods of six fusion algorithms are also discussed. The fusion enhances algorithm＇s performance and expands application＇s scope. Both the method of selecting algorithms and the further research direction in computational intelligence are given through comparison. The simulation study indicates that this algorithm fusion mentality is correct.
出处
《中国电子科学研究院学报》
2007年第1期52-56,61,共6页
Journal of China Academy of Electronics and Information Technology
关键词
神经网络
模糊系统
遗传算法
免疫系统
计算智能
neural network
fuzzy system
genetic algorithm
immune system
computational intelligence
