用RBF神经网络确定上海股市的分形维数

用RBF神经网络确定上海股市的分形维数 The Fractal Dimension of Shanghai Stock Market Based on RBF Neural Networks 徐绪松 XU Xu-song1 1 2 3 first-author 熊保平 XIONG Bao-ping1 1 2 3 龙虎 LONG Hu2 1 2 3 武汉大学商学院技术经济及管理研究所 武汉大学商学院技术经济及管理研究所 Institute of Technology Economics and Management, Business School, Wuhan University Institute of Technology Economics and Management, Business School, Wuhan University 华中科技大学管理学院,湖北,武汉,430072 华中科技大学管理学院,湖北,武汉,430072 Management School, Huazhong University of Science and Technology Management School, Huazhong University of Science and Technology ,湖北,武汉,430074 ,湖北,武汉,430074 从预测能力的角度采用径向基函数 （radialbasisfunction ,简称RBF）神经网络方法计算我国上海股票市场的分形维数 ,并通过RBF神经网络的实验 ,得到上海股市的最小嵌入维数为 6 ,验证了股市分形维数在 2～ 3之间 ,从而进一步确定了我国上海股票市场是一个具有混沌现象的系统 .最后探讨了利用股票市场的混沌特性进行短期预测的效果和可行性 . This paper tests the fractal dimension of Shanghai stock market through the neural networks from the point view of forecasting. Through a data experiment of radial basis function neural network, we find that the optimal embedding dimension is 6. As a result the fractal dimension of Shanghai stock market is about in the interval between 2 and 3. Therefore, the stock market is a chaotic system, and we must take the nonlinear analytical instruments. Finally, we discuss the effect and the reliability of the stock prediction based on chaotic characteristic. 股票市场 混沌 RBF神经网络 分形维数 stock market chaos RBF neural networks fractal dimension F830.9 国家教育部博士点基金资助项目 （ 0 1JB63 0 0 0 9） 2003-03-01 2021-04-01 3