Prediction of multivariable chaotic time series using optimized extreme learning machine
- 1. Center for Control & Intelligence Technology, Nanjing University of Posts and Telecommunications, Nanjing 210003, China;
- 2. School of Iniformation Science & Technology Jiujiang University, Jiujiang 332005, China
- Received Date:
- Accepted Date:
- Available Online:
Project supported by the National Natural Science Foundation of China (Grant No. 60874091), the Six Projects Sponsoring Talent Summits of Jiangsu Province, China (Grant No. SJ209006), the Research Fund for the Doctoral Program of Higher Education of China(Grant No. 20103223110003), the Natural Science Basic Research Project for Universities of Jiangsu Province, China (Grant No. 08KJD510022), the Natural Science Foundation of Jiangsu Province, China (Grant No. BK2010526), the Project for Introduced Talent in Nanjing University of Posts and Telecommunications, China (Grant No. NY209021), and the Scientific Research Innovation Program for the Graduat Students in Jiangsu Province, China (Grant No. CXZZ11 0400).
Abstract: A prediction algorithm of multivariable chaotic time series is proposed based on optimized extreme learning machine (ELM). In this algorithm, a presented composite chaos system and mutative scale chaos method are utilized first to search and optimize the parameters of ELM for improving the generalization performance. Then the optimized ELM is used to predict the multivariable chaotic time series of Rossler coupled system for single step and muti-step, and the scheme is compared with the congeneric method, which shows the validity and stronger ability against noise of the developed algorithm. Finally, the relation between prediction result and number of hidden neurons is discussed.