动态灰预测模型的缓冲适应性建模方法  

Buffered adaptive grey prediction model and its dynamic modelling method

作　　者：徐宁 公彦德 柏菊 XU Ning;GONG Yande;BAI Ju(School of Business,Nanjing Audit University,Nanjing 211815,China;School of Information Engineering,Nanjing Audit University,Nanjing 211815,China)

机构地区：[1]南京审计大学商学院,南京211815 [2]南京审计大学信息工程学院,南京211815

出　　处：《系统工程理论与实践》2019年第10期2619-2627,共9页Systems Engineering-Theory & Practice

基　　金：国家自然科学基金(71701101,71901191)~~

摘　　要：提升对数据特征的适应能力是预测建模的关键问题之一.本文融合缓冲算子方法与灰色滚动预测模型构建适应性灰预测模型,即BARGM模型.该方法将原始序列拆分为连续的数据片段,利用缓冲算子和数据片段上的反馈信息调整变权系数.并用GM(1,1)的衍生模型对调整后的片段数据进行逐步建模和外推;缓冲适应性模型相对传统建模方法具有两个特点,即改变灰模型响应式形式的单一性、具有较强的智能化拓展能力:案例研究采用我国能源相关的温室气体排放数据进行建模测试与对比,建模结果显示拟合精度和预测精度均有明显提升,印证了缓冲适应性建模方法的有效性.A key problem in prediction theory is to improve adaptiveness to data characteristics.This paper proposed a novel method called buffered adaptive grey rolling model or BARGM model,which combines the buffer operator and grey rolling model.First,the new model decomposes original sequence into continuous data slices,and uses feedback input to adjust modeling data by buffer operator with variable weight on each data slice,then a extended grey forecasting model is utilized to model the adjusted data slices stepwisely and generate the extrapolation values.With comparison to GM(1,1),BARGM has two obvious features:Changing the monotonicity of traditional time response function and strong expansibility.The case study uses the BARGM and benchmark models to analyze China’s greenhouse gas emission data,and the results demonstrate that the accuracy of fitting and prediction improve obviously,and prove the new model’s effectiveness.

关 键 词：灰色预测 滚动建模 缓冲算子 时间响应 

分 类 号：N941.5[自然科学总论—系统科学]