摘要
分析了基于数据和基于模型的两类工控系统生产过程异常检测方法的优缺点。针对化工生产过程中工况参数变量之间的变化随工艺装置与生产条件而变化的实际,提出基于工艺生产模型的特征变量模型,并结合模糊C均值聚类算法对化工生产过程异常数据进行检测,基于TE模型的4个场景的仿真结果验证了该方法的有效性。
Based on the data and model,both advantages and disadvantages of detection methods for two industrial control system's abnormal operation were analyzed.In view of the fact that the variation of working condition parameter variables varies with process equipment and production conditions in chemical production,a characteristic variable model based on process model was proposed,including having fuzzy C-means clustering algorithm adopted to detect abnormal data in chemical production.The simulation results of four scenes based on TE model verify effectiveness of the method proposed.
作者
李俊杰
马军鹏
马春雷
贺海波
朱天宇
陈吉
王广平
崔鹏
曹彦东
LI Jun-jie;MA Jun-peng;MA Chun-lei;HE Hai-bo;ZHU Tian-yu;CHEN Ji;WANG Guang-ping;CUI Peng;CAO Yan-dong(China Coal Shaanxi Energy and Chemical Industry Group Co.,Ltd.)
出处
《化工自动化及仪表》
CAS
2024年第6期1028-1034,共7页
Control and Instruments in Chemical Industry
关键词
异常数据检测
工控系统安全
模型分析
数据挖掘
聚类检测
small target defect detection
YOLOv7-tiny
multi-angle data augmentation
feature extraction
attention mechanism
loss function
