基于BP神经网络的股指预测系统

发布时间：2018-04-14 04:14

本文选题：人工神经网络 + BP算法　；参考：《大连理工大学》2012年硕士论文

【摘要】：股票市场自诞生以来,随着经济的发展,在金融市场中占据着非常重要的地位。但是股市受到国家政策、经济环境、突发事件和人为操控等诸多因素的影响,使投资者在享受高收益的同时也承担着巨大的风险。对于占股市大多数的中小投资者来说,如果可以预测股市的走势,那么就可以尽可能地实现收益最大化和风险最小化。在股票市场中,股市指数体现着股市的整体走势,因此对股指进行预测不仅具有理论研究意义,更具有重要的现实意义。通过对股市预测的分析,了解到股市是一个复杂的非线性动态系统,传统的线性预测方法的预测效果不甚理想,而BP神经网络具有很强的非线性逼近能力、自学习能力和自适应能力,由于具有这些优点,BP神经网络成为股市预测领域使用广泛的方法之一。本文对基于BP神经网络的股指预测系统进行了功能性需求和非功能性需求的分析,在需求分析的基础上,根据系统需求提出的设计目标和原则,进行了系统的架构设计和功能设计,然后在详细设计阶段,对各模块进行了详细的功能设计和数据库设计。系统的核心模块为股指预测模块,本文重点对BP神经网络的模型设计、参数选取进行了实验分析,并结合优化网络拓扑结构、加入动量项和改进激活函数来对BP神经网络进行改进,解决BP神经网络预测精度不高、收敛速度慢和不稳定等问题,以提高系统的性能和预测准确率。最后通过预测结果表明,本文实现的基于BP神经网络的股指预测系统具有收敛速度快和预测精度高的优点,具有很高的实用价值。但是该预测系统还存在着一些缺点,如预测范围单一,只进行了上证综合指数的预测；预测精度尚有提高的空间以及系统的实用性尚待进一步的验证。本文的下一阶段工作是研究如何将使系统的应用范围更加广泛以及如何结合其他算法对BP神经网络预测模型进行改进,以获得更好的预测结果。
[Abstract]:Since the birth of the stock market, with the development of economy, it occupies a very important position in the financial market.However, the stock market is affected by many factors, such as national policy, economic environment, unexpected events and artificial manipulation, which make investors enjoy high returns and bear huge risks at the same time.For the small and medium-sized investors who make up the majority of the stock market, if they can predict the trend of the stock market, they can maximize the return and minimize the risk as far as possible.In the stock market, the stock market index reflects the overall trend of the stock market, so the prediction of the stock index is not only of theoretical significance, but also of practical significance.Through the analysis of the stock market forecast, it is found that the stock market is a complex nonlinear dynamic system, the traditional linear forecasting method is not very good, and the BP neural network has a strong nonlinear approximation ability.Self-learning ability and adaptive ability, because of these advantages, BP neural network has become one of the widely used methods in the field of stock market prediction.This paper analyzes the functional requirements and non-functional requirements of the stock index forecasting system based on BP neural network. On the basis of the demand analysis, the design objectives and principles are put forward according to the requirements of the system.The architecture and function of the system are designed, and then in the detailed design stage, the detailed functional design and database design of each module are carried out.The core module of the system is stock index prediction module. This paper focuses on the model design and parameter selection of BP neural network, and combines with the optimization of network topology.In order to improve the performance and prediction accuracy of BP neural network, momentum term and activation function are added to solve the problems of low prediction accuracy, slow convergence rate and instability of BP neural network.Finally, the prediction results show that the stock index prediction system based on BP neural network has the advantages of fast convergence and high prediction accuracy, and has high practical value.However, there are still some shortcomings in the prediction system, such as the single range of prediction, only the prediction of the Shanghai Composite Index, the room for improvement of the prediction accuracy and the practicability of the system need to be further verified.The next stage of this paper is to study how to make the application of the system more extensive and how to improve the BP neural network prediction model combined with other algorithms in order to obtain better prediction results.
【学位授予单位】：大连理工大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP183;F832.5

【参考文献】