模糊认知图在时间序列预测中的应用综述  

Review of the Application of Fuzzy Cognitive Maps in Time Series Prediction

作　　者：秦墩旺 吴立锋[1,2] QIN Dun-wang;WU Li-feng(Information Engineering College,Capital Normal University,Beijing 100048,China;Beijing Key Laboratory of Electronic System Reliability Technology,Capital Normal University,Beijing 100048,China)

机构地区：[1]首都师范大学信息工程学院,北京100048 [2]首都师范大学电子系统可靠性技术北京重点实验室,北京100048

出　　处：《小型微型计算机系统》2023年第10期2314-2322,共9页Journal of Chinese Computer Systems

基　　金：北京市自然科学基金项目(L211018)资助.

摘　　要：时间序列预测是基于当前及历史数据对未来演化趋势的推演.准确的、可解释的时间序列预测是进行科学决策的关键技术支撑,广泛应用于金融、交通、气象等诸多领域.具有可解释性和强推理能力的模糊认知图已在时间序列预测中取得较好的效果,但目前尚无文献对该方法进行全面综述.为此,本文首先对模糊认知图及扩展的高阶模糊认知图、直觉模糊认知图和深度模糊认知图进行梳理,并在此基础上归纳了学习模糊认知图的优化算法.其次,具体介绍了模糊认知图以及扩展的模糊认知图在时间序列预测中的应用,并做出系统性的总结.最后,对模糊认知图在时间序列预测中的发展趋势进行展望.Time series forecasting is the extrapolation of future evolutionary tendencies based on current and historical data.Accurate and interpretable time series forecasting is a key technical support for scientific decision making and is widely used in many fields such as finance,transportation and meteorology.Fuzzy cognitive maps with interpretability and strong inference ability have achieved good results in time series forecasting,but there is no comprehensive review of this method in the literature.To this end,this paper firstly outlines fuzzy cognitive maps and the extended higher-order fuzzy cognitive maps,intuitionistic fuzzy cognitive maps together with deep fuzzy cognitive maps,and on this basis summarizes the optimization algorithms for learning fuzzy cognitive maps.Secondly,the applications of fuzzy cognitive maps and their extensions in time series forecasting are specifically introduced,followed by a systematic summary.Finally,the development trend of fuzzy cognitive maps in time series forecasting is prospected.

关 键 词：模糊认知图 可解释性 优化算法 时间序列预测 

分 类 号：TP389[自动化与计算机技术—计算机系统结构]