过程工业大数据建模研究展望

发布时间：2018-07-24 08:13

【摘要】：人们对大数据的认识已从"3Vs"(Volume—大容量;Variety—多样性;Velocity—处理实时性)、"4Vs"("3Vs"与Value—价值)、到现今的"5Vs"("4Vs"与Veracity—真实性).在此背景下,首先分析过程工业大数据的"5Vs"特性;接下来,综述现有数据建模方法,并结合过程工业大数据特有性质(包括:多层面不规则采样性、多时空时间序列性、不真实数据混杂性)论述现有数据建模方法应用于工业大数据建模时的局限;最后,探讨过程工业大数据建模有待研究的问题,包括:1)多层面不规则采样数据的潜结构建模;2)用于事件发现、决策和因果分析的多时空时间序列数据建模;3)含有不真实数据的鲁棒建模;4)支持实时建模的大容量数据计算架构与方法.
[Abstract]:Big data has been recognized from "3Vs" (Volume-large capacity Variety- diversity), "4Vs" ("3Vs" and Value- value) to today's "5Vs" ("4Vs" and Veracity- authenticity). In this context, the "5Vs" characteristics of process industry big data are analyzed first, and then the existing data modeling methods are reviewed, and combined with the unique properties of process industry big data (including multi-level irregular sampling, multi-spatio-temporal time series, etc.) The limitations of existing data modeling methods in industrial big data modeling are discussed. Finally, the problems to be studied in process industry big data modeling are discussed. Including: 1) modeling of subsurface structures of multi-layer irregular sampling data 2) for event detection, Multi-spatio-temporal time series data modeling for decision and causality analysis (3) robust modeling with unreal data 4) large capacity data computing architecture and method supporting real-time modeling.
【作者单位】：东北大学流程工业综合自动化国家重点实验室;美国南加州大学化工系;香港中文大学(深圳);
【基金】：国家自然科学基金(61304107;61490704;61573022;61290323;61203102) 中国博士后科学基金(2013M541242) 博士后国际交流计划派出项目(20130020) 中央高校基本科研业务费(N130408002;N130108001)资助~~
【分类号】：TP311.13

【相似文献】