[1]赵克勤.集对分析在系统智能预测中的应用综述[J].智能系统学报,2022,17(2):233-247.[doi:10.11992/tis.202103023]
ZHAO Keqin.Application overview of set pair analysis in intelligent prediction system[J].CAAI Transactions on Intelligent Systems,2022,17(2):233-247.[doi:10.11992/tis.202103023]
点击复制
《智能系统学报》[ISSN 1673-4785/CN 23-1538/TP] 卷:
17
期数:
2022年第2期
页码:
233-247
栏目:
综述
出版日期:
2022-03-05
- Title:
-
Application overview of set pair analysis in intelligent prediction system
- 作者:
-
赵克勤
-
诸暨市联系数学研究所,浙江 诸暨 311800
- Author(s):
-
ZHAO Keqin
-
Institution of Zhuji Connection Mathematics, Zhuji 311800, China
-
- 关键词:
-
集对分析; 系统智能预测; 预测模型; 数据结构; 聚类; 动态优化; 联系数; 不确定性分析; 信息能
- Keywords:
-
set pair analysis; system intelligent prediction; prediction model; data structure; cluster; dynamic optimization; connection number; uncertainty analysis; information ability
- 分类号:
-
TP311
- DOI:
-
10.11992/tis.202103023
- 摘要:
-
凡事预则立,不预则废。但事物的预测面临不确定性干扰。本文综述集对分析理论在天气降水预报、沙尘暴预报、水文水资源和供需水预测、电力与能源预测、地质灾害预测、民航风险与事故预测、作物产量预测、流脑预测、社会经济预测等方面的应用,并把基于集对分析理论的系统智能预测建模基本步骤归纳为3步。首先,构造集对并分析集对中两个集合的全部关系,包括确定的关系和不确定的关系,根据关系的结构选用适当的联系数作为集对的特征函数;第二步,建立基于联系数的预测模型,包括利用联系数改进和完善已有的预测模型; 第三步,利用模型的计算和围绕模型的不确定性分析做出预测或预报,包括回顾性预测和当前场景下的实时预测,其中围绕模型的不确定性分析是关键,由此保证和提高预测精度;从而为不确定性系统的智能预测开辟了一条富有成效的新途径。
- Abstract:
-
Preparedness ensures success, whereas unpreparedness spells failure. However, predictions are subject to uncertainty. Accordingly, this paper summarizes the application of set pair analysis theory (SPAT) in weather and precipitation forecast, sandstorm forecast, hydrology, water resources; water supply; and demand forecast, electric power and energy forecast, geological disaster forecast, civil aviation risk and accident forecast, crop yield forecast, cerebral forecast, and social economy forecast. The basic steps of intelligent prediction system modeling based on SPAT are summarized into three steps: First, a set pair is constructed; all the relationships between two sets in the set pair, including relationships that are defined and not defined, are analyzed; and proper connection numbers are selected as the characteristic function according to the relationships. Second, a prediction model based on connection numbers is established, including improving and perfecting the existing prediction models using the connection numbers. Third, a prediction or forecast is made using the calculation of the model and uncertainty analysis around the model (being the key part), including retrospective and real-time predictions under current scenarios. By doing so, the prediction accuracy is guaranteed and enhanced; thus, an effective new way toward intelligent prediction of uncertain systems is opened.
更新日期/Last Update:
1900-01-01