摘要
针对由于流程工业生产系统具有变量多、维度高、耦合关系复杂等特性导致的异常监测数据检验与修复难题,提出了一种采用压缩感知的流程工业异常监测数据检验与修复方法。压缩感知算法能够使用极少的测量数据重构出原稀疏信号或能够在某稀疏基上得到稀疏表达的信号。结合压缩感知原理及重构成功率指标,构建出重构非线性系统方程的数学模型,该模型仅使用少量的监测数据就可以重构出反映状态感知网络节点动态关系的系统方程。将重构的系统方程代入考虑系统偏差的求解结构中得到监测数据的解析值,对比监测数据的实际值与解析值,若实际值超差则用解析值替代实际值,从而实现异常监测数据的检验与修复。通过某煤化工企业压缩机子系统应用案例验证所提方法的有效性,结果表明:在监测数据正常的情况下,实际数据与修复数据的相对误差波动极小,而当监测数据出现异常时,相对误差发生剧烈变化,证明提出的算法能够检测到异常监测数据,并能够较好地恢复监测数据的原貌。
Aiming at verifying and restoring abnormal monitoring data from the production system of process industry with mufti-variables,high-dimensionality and complex coupling relationship,a verification and restoration method for abnormal monitoring data is proposed by compressive sensing.The compressive sensing algorithm can reconstruct original sparse signal with very few measurement data or obtain the signal of sparse expression on a sparse basis.Combining the compressive sensing algorithm with the success rate of reconstruction,a mathematical model is established,which can reconstruct the system equation reflecting the dynamic relationship of the nodes in state-aware network with only a small number of monitoring data.The analytical value of monitoring data is obtained by substituting the reconstructed system equation into the solving structure considering system deviation.Compared the actual value with the analytical value of monitoring data,if the actual value is overproof,the analytical value is used instead of the actual value to verify and restore abnormal monitoring data.The validity of the proposed method is verified by an application case of compressor subsystem in a coal chemical enterprise.When monitoring data are normal,the relative error between the actual data and the restored data fluctuates slightly;when monitoring data are abnormal,the relative error changes dramatically.
作者
徐光南
高智勇
梁艳杰
高建民
刘倩倩
程亚辉
XU Guangnan;GAO Zhiyong;LIANG Yanjie;GAO Jianmin;LIU Qianqian;CHENG Yahui(State Key Laboratory for Manufacturing Systems Engineering,Xi’an Jiaotong University,Xi’an 710049,China)
出处
《西安交通大学学报》
EI
CAS
CSCD
北大核心
2020年第2期59-70,共12页
Journal of Xi'an Jiaotong University
基金
国家自然科学基金资助项目(51375375)
关键词
压缩感知
流程工业
重构
异常数据检验
数据修复
compressive sensing
process industry
reconstruct
abnormal data verification
data restoration
