灰色微分动态多变量预测模型及其应用  

Grey differential dynamic multivariate forecasting model and its application

作　　者：段辉明 何成琳 王思琦 黄江波 DUAN Huiming;HE Chenglin;WANG Siqi;HUANG Jiangbo(School of Science,Chongqing University of Posts and Telecommunications,Chongqing 400065,China)

机构地区：[1]重庆邮电大学理学院,重庆400065

出　　处：《系统工程理论与实践》2022年第5期1402-1412,共11页Systems Engineering-Theory & Practice

基　　金：重庆社科规划博士项目(2020BS58);国家自然科学基金面上项目(72171031,71871174)。

摘　　要：针对数据驱动建模方法在表征系统特性时的不足,提出了灰色微分动态多变量预测模型.新模型将系统特性的行为序列与影响序列用于建模,增强了系统动态性和非线性性;同时运用最小二乘原理获得模型参数估计式,利用拉普拉斯变换推导模型的近似时间响应式.在此基础上,将新模型应用于欧洲货币联盟、中东与北非及撒哈拉以南非洲三个典型地区的碳排放量预测,应用实例表明:灰色微分动态多变量预测模型预测效果优于其它三种经典的灰色预测模型,能有效预测三个地区未来五年的碳排放量.与此同时说明新模型能够更好地描述多因素系统动态性问题,从而有效地提升传统灰色多变量模型的建模精度.To solve the shortcoming of data-driven modeling method in characterizing system characteristics,a grey differential dynamic multivariable prediction model was proposed.In the new model,the behavior sequence and influence sequence with system characteristics are used to model,which strengthens the system dynamics and nonlinearity.At the same time,the least square principle is used to obtain the parameter estimation,and Laplace transform is used to derive the approximate time response.On this basis,the model was applied to carbon emissions projections for three typical regions:The European Monetary Union,the Middle East and North Africa,and sub-Saharan Africa.Application examples show that prediction effect of grey differential dynamic multivariable prediction model is better than other three classic grey prediction model,can effectively predict the carbon emissions of three regional in the next five years.Meanwhile,it shows that the new model can better describe the dynamic problem of multi-factor system,so as to effectively improve the modeling accuracy of the traditional grey multivariable model.

关 键 词：微分动态模型 多变量灰色预测模型 拉普拉斯变换 二氧化碳排放量预测 

分 类 号：N941.5[自然科学总论—系统科学]