BP神经网络和模糊时间序列组合预测模型及其应用  

Composite prediction model of BP neural networks and fuzzy time series and its application

作　　者：石慧[1] 王玉兰[1] 翁福利[2] 

机构地区：[1]成都理工大学管理科学学院,成都610059 [2]电子科技大学数学科学学院,成都611731

出　　处：《计算机应用》2011年第A02期90-91,102,共3页journal of Computer Applications

摘　　要：为了解决非线性的时间序列预测问题,提出了BP神经网络和模糊时间序列相结合的预测模型。利用BP神经网络自学习和模糊集能够更客观反应实际情况,通过对时间序列差分模糊化建立数学模型,BP神经网络进行训练,最后去模糊化还原实际。将这种预测方法应用到矿产资源镍价格中,取得了较好的效果。To solve nonlinear time series prediction problems,a composite model was put forword.The model was based on BP neural networks and fuzzy time series.BP neural networks capture the feature of self-learing.Fuzzy sets can better reflect real situations.The mathematical model was established by difference and fuzzification of time series.BP neural networks trained the model.Finally,the fuzzy model was turned into the real model by elimating fuzzification.The proposed model was used to forecast nickel price,and achived good reselts.

关 键 词：模糊时间序列 BP神经网络 非线性关系 镍价格 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]