Multi-Steps Forecast of Chaotic Sequence Based on IMA and NN
Neural Networks are widely applied in the dynamic forecast of those multiple factors and non-linear complex processes. The Immune Algorithm (IMA) based on vector square is suggested as the learning algorithm. The weights matrix of the network and the threshold matrix of nerve cells are simply combined as an antibody. The NN is firstly trained based on IMA and learning samples, the optimal result is the trained NN. Then the forecast performance of the IMA-NN is tested by unknown samples. Simulation is performed based on MATLAB, results show that IMA-NN can obtain upper precision both in the approaching of learning samples curve and in the forecasting of unknown samples, so it is a more effective method.
vector square Immune Algorithm Neural Network forecast chaotic sequence
JING Xian-yong XIAO Ming-qing LIU Zhan-chen HOU Man-yi
Airforce Engineer University Engineering College, Xian 710038, China The Ordnance Department, Aviation University of Airfore
国际会议
第八届国际测试技术研讨会(8th International Symposium on Test and Measurement)
重庆
英文
1547-1550
2009-08-01（万方平台首次上网日期，不代表论文的发表时间）