基于溷沌神经网络的股票分析及其预测.pdfVIP

19 3 Vol. 19 o. 3 2009 3 COM PUT ER TECH OLOGY A D DEVELOPME T Mar. 2009 1 2 张中华 , 丁华福 ( 1. 哈尔滨理工大学计算机科学与技术学院, 黑龙江哈尔滨 150080; 2. 哈尔滨工业大学计算机科学与技术学院,黑龙江哈尔滨 150001) : , 1990. 12. 19 2008.4.24 ,, 1990 2008, ,Lyapunov , ,, , :; ;;Lyapunov ; :TP183 :A : 1673- 629X(2009)03- 0185- 04 Analysis and Prediction of Stock Based on Chaotic Neural Networks 1 2 ZHA G Zhong-hua ,DI G Hua-fu ( 1.Sch.of Computer and Techn.,Haerbin Univ.of Sci.and Techn. ,Haerbin 150080, China; 2.Sch. of Computer and Techn.,Haerbin Univ. of Industry, Haerbin 150001,China) Abstract:The chaotic dynamics theory has provided an analytic method to stock price volatility in the stock market. To find out whether the price of stock has the chaotic behavior, has analyzednonlinear characterof thestock price, based on thecloseindex of stock, from the 19th of Dec in 1990 to 24th of Apr in 2008, and have reconstructed the strange attractor of the time sequence by the method of recon- structing phase space, have calculated correlated dimension, and have proved that the exponent of Lyapunov is positive, then affirmed that thechaotic behavior does exit in thetime sequence of stock. The neural network is a nonlinear system, it can approach thenonlinear function by learning, by using the neural network, can forecast the stock s direction of thenext day. Key words:chaotic behavior; reconstructed phase space; correlated dimension; Lyapunov exponent; neural network 0 [ 3,4] , Pa