摘要
针对GM(1,1)模型预测精度差的问题,采用马尔可夫链、残差数据正数化等多种方法修正残差,按等维新息的思路建立多个改进型GM(1,1)模型,并提出了基于神经网络实现的改进型灰色组合预测模型及预测算法。仿真分析表明,通过该模型可以寻求到多个改进型GM(1,1)模型预测值的最佳组合。
For the poor prediction precision of GM （1,1） model, several methods to modify remnant difference are adopted, such as Markov Chain or making remnant difference to positive, etc. According to the principles of equal dimension and new information GM （1,1）, several improved GM （1,1） models are founded. Moreover, a combination forecasting model of improved GM （1, 1） and its algorithm based on neural network realization are put forward. Simulation analysis shows that with the model the superior forecasting value can be found, which is the best combination of the forecasting values calculated by the above several improved GM（1,1） models. It is a feasible method to improve the forecasting precision of traditional GM（1,1） model.
出处
《交通与计算机》
2006年第6期9-12,共4页
Computer and Communications
基金
国家自然科学基金项目资助(批准号:60664001)
江西省自然科学基金项目资助(批准号:0511030)
关键词
灰色预测
马尔可夫链
残差
等维新息灰色模型
神经网络
gray predietion
Markov Chain
residual error
equal dimension and new information gray forecasting model
neural network
