| 引用本文: | 谢永芳,李理,谢世文,陈晓方.基于半定量概率图模型的溯因分析方法[J].控制理论与应用,2023,40(3):419~429.[点击复制] |
| XIE Yong-fang,LI Li,XIE Shi-wen,CHEN Xiao-fang.Method of root cause analysis based on semi-quantitative probabilistic graphical model[J].Control Theory and Technology,2023,40(3):419~429.[点击复制] |
|
| 基于半定量概率图模型的溯因分析方法 |
| Method of root cause analysis based on semi-quantitative probabilistic graphical model |
| 摘要点击 1169 全文点击 366 投稿时间：2021-04-29 修订日期：2022-05-01 |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| DOI编号 10.7641/CTA.2022.10368 |
| 2023,40(3):419-429 |
| 中文关键词 铝电解 半定量 概率图模型 不确定性 溯因分析 |
| 英文关键词 aluminum electrolysis semi-quantitative probabilistic graphical model uncertainty root cause analysis |
| 基金项目 广东省重点领域研发计划项目(2021B0101200005), 国家自然科学基金重点项目(62233018), 湖南省自然科学基金杰出青年项目(2023JJ10079), 国 家自然科学基金青年基金项目(62003370)资助. |
|
| 中文摘要 |
| 复杂工业系统的故障原因定位可协助操作人员快速调整设备运行参数, 保障生产高效稳定地运行. 铝电解
过程机理复杂且外部因素干扰多, 信息具有不确定性特征, 难以建立精确的定量模型, 而定性分析的准确度不高.
为此, 本文针对铝电解溯因过程的层次性、相关性、不确定性的特点, 构建了一种基于半定量概率图模型的溯因分
析框架, 将定量和定性分析相结合, 通过不确定理论对信息进行处理和描述, 采用图形符号可视化知识变量间的因
果关系, 再基于概率图模型的推理方法实现不确定性条件下的溯因诊断, 为实现铝电解异常槽况的原因分析与定位
提供了理论支撑. |
| 英文摘要 |
| Cause analysis of complex industrial systems is beneficial for technicians to quickly adjust the operating
parameters, so that the system can work efficiently and stably. As the mechanism of the aluminum electrolysis is complex
and there are many external interferences, the information has the characteristics of uncertainty. It is difficult to establish
an accurate quantitative model, while the accuracy of qualitative analysis is not high. Therefore, a framework of cause
analysis model based on the semi-quantitative probabilistic graph is proposed in the paper, which deals with the problem of
hierarchy, correlation and uncertainty in the aluminum electrolysis traceability process. The proposed model combines the
quantitative and qualitative methods, and applies uncertainty theory to represent knowledge. Then, graphic symbols are used
to visualize the causal relationship between variables, and casual inference is performed by probabilistic graphical model.
Thus, it can provide theoretical support for realizing the cause analysis of abnormal conditions in aluminum electrolysis
production. |
|
|
|
|
|