融合群智能方法BP神经网络模型及其在股市预测中的应用

发布时间：2018-03-02 06:11

本文关键词： BP神经网络粒子群优化算法趋势因子股市预测　出处：《吉林财经大学》2013年硕士论文　论文类型：学位论文

【摘要】：鉴于智能计算方法具有自适应、自学习、并行性等优点，本文以智能计算方法为基础，，通过引入趋势因子和群智能等方法，提出了改进BP神经网络模型并将其应用于股市预测领域中。具体内容包括：（1）通过对BP神经网络的输入层、隐层和输出层节点数目的确定，建立基于BP神经网络的股市预测模型，并将其对深圳成分指数进行仿真模拟预测。实验结果表明，BP神经网络用于股市预测领域是可行的，有效的，具有一定的优越性。（2）鉴于趋势因子具有纠正预测方向的特性，为进一步提高BP神经网络的预测性能，本文将趋势因子引入到BP神经网络中，提出了引入趋势因子BP神经网络，即DF-BPNN网络(BP Neural Network with Direction Factor)。实验结果表明，与基本BP神经网络相比，本文提出的DF-BPNN模型的预测性能优于BP神经网络，其预测精度有进一步的改善。（3）鉴于粒子群优化算法具有较好的全局搜索能力和优化性能，本文利用粒子群优化算法对BP神经网络的权重和阈值进行优化，提出了一种基于群智能的PSO-BP混合神经网络模型即PSO-BP NN（PSO-BP NeuralNetwork）。另外，采用本文提出的PSO-BPNN模型对深圳成分指数股市进行预测，得到了令人满意的结果。（4）针对本文提出的引入趋势因子BP神经网络和PSO-BP混合神经网络，与基本BP神经网络的预测性能进行比较研究。通过实验获得的仿真模拟图与数值结果表明，本文提出的引入趋势因子BP神经网络与PSO-BP混合神经网络的预测性能均优于基本BP神经网络，但引入趋势因子BP神经网络与PSO-BP混合神经网络相比，其预测性能差别较小。
[Abstract]:In view of the advantages of self-adaptive, self-learning and parallelism, the intelligent computing method is based on the introduction of trend factor and swarm intelligence. This paper proposes an improved BP neural network model and applies it to the field of stock market prediction. The content includes: 1) by determining the number of nodes in the input layer, hidden layer and output layer of BP neural network, the stock market forecasting model based on BP neural network is established. The experimental results show that the BP neural network is feasible, effective and has some advantages in the field of stock market forecasting. (2) since the trend factor has the characteristics of correcting the forecast direction, In order to further improve the prediction performance of BP neural network, the trend factor is introduced into BP neural network in this paper, and a BP neural network with trend factor is proposed. The experimental results show that the BP neural network is compared with the basic BP neural network. The prediction performance of the proposed DF-BPNN model is better than that of BP neural network, and its prediction accuracy is further improved. In this paper, particle swarm optimization algorithm is used to optimize the weight and threshold of BP neural network, and a PSO-BP hybrid neural network model based on swarm intelligence is proposed, I. E. PSO-BP NN(PSO-BP neural network. The PSO-BPNN model proposed in this paper is used to predict the Shenzhen component index stock market, and a satisfactory result is obtained. The BP neural network and the PSO-BP hybrid neural network are introduced in this paper, and the trend factor BP neural network and the PSO-BP hybrid neural network are introduced in this paper. The prediction performance of BP neural network is compared with that of basic BP neural network. The simulation and numerical results obtained from experiments show that, The prediction performance of BP neural network and PSO-BP hybrid neural network proposed in this paper is better than that of basic BP neural network, but the difference of prediction performance between BP neural network and PSO-BP hybrid neural network is smaller.
【学位授予单位】：吉林财经大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：F224;F832.5

【参考文献】