大数据下复杂流程工业的分布式过程监测与故障诊断
结题报告
批准号:
61973023
项目类别:
面上项目
资助金额:
60.0 万元
负责人:
王晶
依托单位:
学科分类:
控制理论与技术
结题年份:
2023
批准年份:
2019
项目状态:
已结题
项目参与者:
王晶
国基评审专家1V1指导 中标率高出同行96.8%
结合最新热点,提供专业选题建议
深度指导申报书撰写,确保创新可行
指导项目中标800+,快速提高中标率
客服二维码
微信扫码咨询
中文摘要
现代流程工业的首要任务是安全生产,因此对其进行过程监控和故障诊断对提升智能化安全生产有非常重要的意义。项目以长流程、强耦合、超高维、强非线性的复杂流程工业系统为对象,开展分布式过程监测与故障诊断研究。与传统集中式多元统计监控不同,提出过程分解、局部监测与融合诊断的三层分布式监控架构:过程分解层以监测目标为导向采用优化实验设计方法从超高维过程变量中优化筛选监测变量,并进行合理的数据分块,达到工业大数据降维和简化目的;局部监测层两条路线并行,构建多方法融合的潜结构投影模型,确保对非线性特性的提取,实现子系统的有效监测;同时构建基于概率密度估计的因果图模型,实现连续过程变量的故障传播定量分析;融合诊断层利用各子系统的监控统计量,构建融合诊断指标,实现整个流程的因果推理与故障溯源。项目提出一套适合复杂流程工业的分布式监控与诊断的理论体系与技术方法,其实施有助于提高我国流程工业的智能优化制造水平。
英文摘要
The primary task of modern process industry is safe production, so its process monitoring and fault diagnosis are very important for intelligent manufacturing. This proposal focuses on the distributed monitoring of complex processes with long processes, strong coupling, ultra-high dimensions and strong non-linearity characteristics. Three layer distributed monitoring architecture consists of process decomposition, local monitoring and fusion diagnosis. First, aim to reduce the dimensionality of industrial big data and simplify the complexity of data analysis, the appropriate monitoring variables are designed and selected from the ultra-high dimensional process variables based on the optimal experiment design, then they are decomposed into several subsystem oriented by monitoring target. Then two routes of local monitoring are designed in parallel. A latent structure projection model with multi-method fusion is constructed to extract the nonlinear characteristics and detect the abnormal behavior of subsystems. A causal graph model based on probability density estimation is constructed to realize quantitative analysis of fault propagation of continuous process variables. Finally in the fusion diagnosis layer, fusion diagnosis indicator is given based on the monitoring statistics of each subsystems, then causal reasoning and fault tracing of the whole process is realized. Implementation of all the proposed method will contribute in the statistical monitoring theory of complex industrial process, and advance the intelligent manufacturing level of domestic industry.
期刊论文列表
专著列表
科研奖励列表
会议论文列表
专利列表
DOI:10.3390/s22020422
发表时间:2022-01-06
期刊:Sensors (Basel, Switzerland)
影响因子:--
作者:Zhou M;Zhang Y;Wang J;Shi Y;Puig V
通讯作者:Puig V
DOI:10.3390/fractalfract6050276
发表时间:2022
期刊:Fractal Fract
影响因子:--
作者:Jing Wang;Yi Liu;Haiyan Wu;Shan Lu;Meng Zhou
通讯作者:Meng Zhou
DOI:10.1007/s11432-018-9807-7
发表时间:2021
期刊:Science China. Information Sciences
影响因子:--
作者:Jia Ruixue;Wang Jing;Zhou Jinglin
通讯作者:Zhou Jinglin
DOI:10.1109/tase.2020.3022924
发表时间:2021-10
期刊:IEEE Transactions on Automation Science and Engineering
影响因子:5.6
作者:Xiaolu Chen;Jing Wang;S. Ding
通讯作者:Xiaolu Chen;Jing Wang;S. Ding
DOI:10.1016/j.isatra.2022.07.017
发表时间:2022-07
期刊:ISA transactions
影响因子:7.3
作者:Jing Wang;Pengyang Liu;Shan Lu;Meng Zhou;Xiaolu Chen
通讯作者:Jing Wang;Pengyang Liu;Shan Lu;Meng Zhou;Xiaolu Chen
面向流程工业大数据的轻量级鲁棒学习与故障诊断
  • 批准号:
    62373005
  • 项目类别:
    面上项目
  • 资助金额:
    50.00万元
  • 批准年份:
    2023
  • 负责人:
    王晶
  • 依托单位:
国内基金
海外基金