基于混沌时间序列及弹性反馈算法的股票预测方法研究

发布时间：2018-08-31 19:13

【摘要】：随着中国经济的迅猛发展，理财的概念逐渐在大众心理建立起来，而股票就是一直受大众青睐的理财产品。股票是市场经济融资的重要手段之一，股市的发展不仅体现国家经济的发展，更是关乎千家万户的切身利益。大自然的混沌现象无处不在，由大自然的产物——人所一手操办的股市必然也是一个混沌的系统。因此，本文将通过混沌时间序列及弹性反馈神经网络对股票价格走势进行预测，具体内容安排如下：首先，介绍混沌动力学以及混沌时间序列的相关理论。先介绍混沌的现象、混沌的定义、混沌的基本特性、李雅普诺夫指数等基础的混沌学知识。紧接着简要介绍了混沌时间序列的知识，重点包括相空间重构技术、时间延迟和嵌入维数的确定以及最大李雅普诺夫指数预测方法。其次，介绍了神经网络的基础知识，详细的介绍了反馈神经网络的原理，重点介绍了反馈神经网络的主要算法并比较他们的优缺点，最终论证了为何选择弹性神经网络算法作为本文的预测方法。最后，利用混沌时间序列和弹性反馈神经网络结合的方法对某只股票数据进行预测分析，将预测结果与最大李雅普诺夫指数预测结果及经典反馈神经网络预测结果进行比较。研究表明，结合混沌时间序列和弹性反馈算法对股票进行预测，无论在精度还是性能上都取得了更好的效果。
[Abstract]:With the rapid development of Chinese economy, the concept of financial management is gradually established in the popular psychology, and the stock is always favored by the masses. Stock is one of the important means of market economy financing. The development of stock market not only reflects the development of national economy, but also relates to the vital interests of thousands of households. The chaos of nature is everywhere, and the stock market run by man is a chaotic system. Therefore, in this paper, the stock price trend is predicted by chaotic time series and elastic feedback neural network. The main contents are as follows: firstly, the chaotic dynamics and chaotic time series theory are introduced. This paper first introduces the phenomena of chaos, the definition of chaos, the basic characteristics of chaos, and the basic knowledge of chaos such as Lyapunov exponent. Then, the knowledge of chaotic time series is briefly introduced, including the reconstruction of phase space, the determination of time delay and embedding dimension, and the prediction method of maximum Lyapunov exponent. Secondly, the basic knowledge of neural network is introduced, the principle of feedback neural network is introduced in detail, the main algorithms of feedback neural network are introduced, and their advantages and disadvantages are compared. Finally, the paper demonstrates why the elastic neural network algorithm is chosen as the prediction method in this paper. Finally, the method of combining chaotic time series with elastic feedback neural network is used to predict and analyze the stock data. The prediction results are compared with those of the largest Lyapunov exponent and the classical feedback neural network. The results show that the accuracy and performance of stock prediction are improved by using chaotic time series and elastic feedback algorithm.
【学位授予单位】：南京航空航天大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：F832.51;F224

【参考文献】