季节性弱化缓冲算子的构造与应用  

The construction and application of seasonal weakening buffer operator

作　　者：何凌阳 王正新 HE Lingyang;WANG Zhengxin(School of Economics,Zhejiang University of Finance&Economics,Hangzhou 310018,China)

机构地区：[1]浙江财经大学经济学院,杭州310018

出　　处：《系统工程理论与实践》2022年第1期13-23,共11页Systems Engineering-Theory & Practice

基　　金：国家自然科学基金(71971194,71571157)。

摘　　要：针对现有弱化缓冲算子不能有效处理季节性冲击扰动系统的建模预测问题,本文构造了两类季节性弱化缓冲算子,分别为季节性平均弱化缓冲算子和季节性全信息变权弱化缓冲算子.在此基础上,本文进一步探讨了季节性弱化缓冲算子的缓冲强度与光滑性,发现二者均优于该季节算子所对应的经典缓冲算子.最后,本文基于灰狼算法给出了季节性全信息变权缓冲算子的权重优化方案,并以第二产业季度增加值数据为例证实了季节性缓冲算子的有效性.预测结果表明:对于受冲击扰动的季节性时序数据,本文提出的季节性缓冲算子的适应能力和预测精度显著优于经典缓冲算子;进一步与SARIMA和EMD-ARIMA模型的比较发现,两类季节性缓冲算子的预测精度与EMD-ARIMA模型相当,三者的平均预测相对误差均在3%左右,而SARIMA模型的平均预测误差则高达15.65%.Given that the existing weakening buffer operators cannot effectively deal with the modeling and forecasting problems of seasonal shock disturbance systems,this paper constructs two types of seasonal weakening buffer operators,that is,seasonal average weakening buffer operator and seasonal fullinformation variable weight weakening buffer operator.On this basis,this paper further discusses the buffer strength and smoothness of the seasonal weakening buffer operator,and finds that they are better than that of the corresponding classical one.Finally,based on the grey wolf algorithm,this paper gives the weight optimization scheme of seasonal full-information variable weight weakening buffer operator and takes the impact and disturbance data prediction of the quarterly added value data of the secondary industry as an example to verify the effectiveness of the seasonal buffer operator.The forecasting results show that:For seasonal time series influenced by disturbance items,the adaptability and prediction accuracy of the seasonal buffer operator proposed in this paper is significantly better than that of the classic buffer operator.Further findings show that,compared with the SARIMA and EMD-ARIMA models,the prediction accuracy of two types of seasonal buffer operators is equivalent to that of the EMD-ARIMA model:Their average prediction relative error is about 3%,while that of the SARIMA model is as high as 15.65%.

关 键 词：灰色系统 季节性扰动 缓冲算子 季节因子 预测 

分 类 号：N941.5[自然科学总论—系统科学]