摘要
局外点检测对于注重试验和数据采集的化学化工领域,其重要性不可忽视。结合化学化工的数据特点,对局外点检测方法的理论基础、应用需求、研究思路、技术关键和发展进程进行了综合分析。分别针对线性和非线性参变对象,概要介绍了当前适用的检测方法,包括各类判别标准、直接和间接方式、单个和多个局外点检测、经典和现代方法以及它们的优缺点和最新进展。最后展望了局外点检测方法的前景和发展方向。
The detection of outlier is very important in chemistry and chemical engineering which emphasizes experimentation and data acquisition. The theoretical foundation, application demand, research approach, critical technology and development of the detection of outlier are analyzed comprehensively, combined with the characteristic of chemical data. With respect to the linear and non-linear system, the current detection solution is introduced, including detection criteria, direct and indirect methods, single and non-single detection, classic and modern means, their advantages and disadvantages and the latest development. The prospect and development direction of the detection of outlier are presented.
出处
《化工进展》
EI
CAS
CSCD
北大核心
2006年第8期913-917,共5页
Chemical Industry and Engineering Progress
基金
国家自然科学基金资助项目(No.20276063)
浙江省重点科技项目(No.2004C21SA120002)
关键词
局外点检测
残差
干净集
稳健回归
神经网络
outlier detection
residuals
clean-set
robust-regression
neural network
