|Table of Contents|

[1] Wen Xiulan, **, Song Aiguo, Duan Jianghai, et al. Evolving Neural Networks Using an Improved Genetic Algorithm* [J]. Journal of Southeast University (English Edition), 2002, 18 (4): 367-369. [doi:10.3969/j.issn.1003-7985.2002.04.016]
Copy

Evolving Neural Networks Using an Improved Genetic Algorithm*()
基于改进遗传算法进化神经网络
Share：

Journal of Southeast University (English Edition)[ISSN:1003-7985/CN:32-1325/N]

Volumn:
18
Issue:
2002 4
Page:
367-369
Research Field:
Information and Communication Engineering
Publishing date:
2002-12-30

Info

Title:
Evolving Neural Networks Using an Improved Genetic Algorithm*
基于改进遗传算法进化神经网络
Author(s):
Wen Xiulan1;  2**;  Song Aiguo1;  Duan Jianghai1;  Wang Yiqing1
1Department of Instrument Science and Technology, Southeast University, Nanjing 210096, China
2Mechanical Engineering College, Inner Mongolia Polytechnic University, Huhot 010062, China
温秀兰1;  2;  宋爱国1;  段江海1;  王一清1
1东南大学仪器科学与工程系, 南京210096; 2内蒙古工业大学机械学院, 呼和浩特 010062
Keywords:
genetic algorithms;  neural network;  nonlinear forecasting
遗传算法;  神经网络;  非线性预测
PACS:
TN911
DOI:
10.3969/j.issn.1003-7985.2002.04.016
Abstract:
A novel real-coded improved genetic algorithm(GA)of training feed-forward neural network is proposed to realize nonlinear system forecast. The improved GA employs a generation-alternation model based the minimal generation gap(MGP)and blend crossover operators(BLX-α). Compared with traditional GA implemented in binary number, the processing time of the improved GA is faster because coding and decoding are unnecessary. In addition, it needn’t set parameters such as the probability value of crossover and mutation by experiences. Therefore, it has the advantages of simple algorithms, strong robustness and high optimization efficiency. Then forecasting nonlinear system using feed-forward neural network is presented. Simulation shows the method is rapid and effective.
本文提出一种新颖的基于实数编码的改进遗传算法用于前馈神经网络的训练, 进而实现对非线性系统预测.该改进遗传算法采用基于代沟最小的代选择模型, 选用BLX-α混合交叉算子.与经典的基于二进制编码的遗传算法相比较, 该算法不需要编码和解码, 所以计算速度快;且不需要根据经验设置交叉和变异概率, 因而算法简单、鲁棒性强、优化效率高.同时给出了应用该算法对前馈神经网络进化时的计算流程.仿真结果证实该方法对非线性系统进行预测是快速有效的.

References:

[1] Lam H K, Ling S H, Leung F H F, et al. Tuning of the structure and parameters of neural network using an improved genetic algorithm[A]. In: Proc of IECON[C]. 2001.25-30.
[2] Holland J H. Adaptation in natural and artificial system[M]. Ann Arbor, MI: University of Michigan Press, 1975.
[3] Satoh H, Yamamura M, Kobayashi S. Minimal generation gap model for gas considering both exploration and exploitation[A]. In: Proc of IIZUKA[C]. 1996. 494-497.
[4] Eshelman L J, Schaffer J D. Real-coded genetic algorithms and interval-schemata[J]. Foundations of Genetic Algorithms, 1993, 2: 187-202.
[5] Ono I, Kobayashi S. A real-coded genetic algorithm for function optimization using unimodal normal distribution crossover[A]. In: Proc 7th ICGA[C]. 1997.246-253.
[6] Gong D, Xu S, Sun X. Research on fast training algorithm for recurrent neural network[A]. In: IEEE Inter Symp on Industrial Electronics[C]. 2001.446-448.

Memo

Memo:
* The project supported by the Natural Science Foundation of Jiangsu Province(BK2001402); Dissertation Foundation of Southeast University - Nanrui Jibao Corporation.
** Born in 1966, female, associate professor.
Last Update: 2002-12-20