摘要
以大型火电机组为研究对象,提出了一种基于互信息(MI)和慢特性分析(SFA)的异常工况检测方法,用于提高工业过程中异常工况检测的准确率和效率。首先,根据过程变量和故障标签的MI值选取高于设定阈值的过程变量;然后,利用慢特征算法提取出特征矩阵,使用两种新的指标计算统计量,通过潜在变量模型的慢特征来检测过程数据的异常;最后,将该方法应用于汽轮机和引风机异常工况案例中,与传统算法的对比实验分析表明该方法有较强的工程应用价值。
To detection the abnormal condition of thermal power plant, the present work proposes a new method based on slow feature analysis(SFA) and mutual information(MI), which could improve the accuracy and efficiency of abnormal condition detection in industrial processes. Firstly, the process variables that are higher than the threshold are selected based on the MI values of process variables and fault labels. Secondly,the characteristic matrix is extracted by the slow feature algorithm, two new indicators are used to calculate the statistics, and the anomaly of the process data is detected by the slow features of the latent variable model.Finally, this method is applied to two industrial cases, steam turbines and induced draft fans. Compared with the traditional algorithm, the analysis shows that this method has strong engineering application value.
作者
王小邦
贺凯迅
苏照阳
WANG Xiao-bang;HEKai-xun;SU Zhao-yang(College of Electrical Engineering and Automation,Shandong University of Science and Technology,Qingdao 266590,China;Key Laboratory of Smart Manufacturing in Energy Chemical Process,Ministry of Education,East China University of Science and Technology,Shanghai,200237,China)
出处
《控制工程》
CSCD
北大核心
2022年第10期1881-1886,共6页
Control Engineering of China
基金
国家自然科学基金资助项目(61803234,61751307)
中央高校基本科研业务费重点科研基地创新基金资助项目。
关键词
异常工况
慢特性分析
互信息
火力发电
特征提取
Abnormal condition
slow feature analysis
mutual information
thermal power
feature extraction
