摘要
针对GM(1,1)预测模型的不足之处,首先基于X(0)序列的相对误差平方和最小的思路,提出了一种新的优化时间响应函数即确定指数函数exp(-at)系数C的方法;第1步采用优化背景值方法确定a,b后,第2步用本文方法确定系数C,得到了一个优化组合的新GM(1,1)预测模型.经大量的数据模拟发现,此优化组合新模型无论对高增长系数,还是对低增长系数都具有极高的模拟与预测精度.
Aiming at the disadvantage of GM （ 1,1 ） forecasting model, this paper firstly utilizes a train of thought that the sum of relative error square function of sequence X^（0） is the smallest,puts forward a new optimal time response function, that is a method to define the coefficient of exponential function e^ -at ; Then, after deciding a, b through the method of optimizing background value in document at the first step, the second one uses our new meth- od to decide the coefficient c, yields a new optimal combination GM （ 1,1 ） forecasting model. The simulation of a great deal of data finds that the new optimal combination model has extremely. The simulation of a great deal of data finds that the new optimal combination model has extremely high simulating and forecasting precision both to high and low developing coefficient.
出处
《西华师范大学学报（自然科学版）》
2007年第1期40-44,共5页
Journal of China West Normal University(Natural Sciences)
基金
四川省教育厅重点科研课题基金资助项目(2006A077)
