基于广义动态灰关联度窗口的时滞分析模型

基于静态最大关联度准则的经典灰关联时滞分析方法对样本时序的依赖性较强,分析结论的代表性较差、甚至会出现自相矛盾的地方.为此,本文探讨了一种新的全局性动态时滞分析方法.首先,设计了新信息优先加权算子,将其引入广义灰色关联度模型,强调新信息在关联分析中的重要性.其次,针对经典的基于灰关联度的时滞分析方法的不足,提出动态灰关联度窗口的概念,对样本时序中所有潜在时滞取值展开系统化的分析.再者,定义了基于动态灰关联度窗口概念的时滞关联度矩阵以及相应的时滞灰关联度向量,设计出灰栈矩阵这一特殊灰关联数据存储结构,动态对比所有潜在时滞关系,找出代表性时滞值.最后,对宏观经济指标之间的时滞关系展开案例分析,并与经典灰关联时滞分析方法进行对比,结果表明本文设计的时滞分析方法具有更好的说服力.

Time delay analysis model based on the generalized dynamic grey incidence window concept

Based on the criterion of the static maximum grey incidence degree, the classic time-delay analysis method strongly depends on time series samples. This affects the representativeness of the delay values, and even causes contradictory results. To solve these problems, this study develops a new dynamic time-delay analysis method. First, a new generalized grey incidence model is constructed by introducing a novel new information priority weighting operator. Then, to overcome the disadvantages of the widely used classic grey incidence-based time delay analysis method, a new concept of dynamic grey incidence window is proposed, which helps to extract all possible time delays between sequences. Based on this new model and the dynamic grey incidence window concept, the time-delay incidence matrix and related time-delay incidence vectors are obtained. A novel grey stack matrix is designed to realize the effective extraction of the representative value from all potential time delays between sequences. Finally, a case study of the time delays between some important macroeconomic indicator sequences is carried out. Compared with the classic method, result shows that the proposed time delay analysis method can provide more reliable representative time delay values.