基于数据驱动的TE及FPSO生产流程故障诊断的研究

发布时间：2018-01-15 11:52

本文关键词：基于数据驱动的TE及FPSO生产流程故障诊断的研究　出处：《天津理工大学》2012年硕士论文　论文类型：学位论文

【摘要】：生产过程的故障诊断一直以来都是流程工业所重视的关键问题之一。深入开展工业过程故障诊断的研究和应用对保障运行安全和提高经济效益均有较大的促进作用。基于数据驱动的故障诊断研究是现在流程工业的一个重要关注点。本文分析了故障诊断的方法和故障诊断发展趋势，通过对主元分析（PrincipleComponent Analysis，PCA）故障诊断法原理以及Hotelling T2统计量、SPE统计量的分析，给出了其故障诊断的过程。同时，分析了几种常用核函数的特点以及核主元分析（Kernel Principle ComponentAnalysis，，KPCA）故障检测算法的原理和及其应用过程。对田纳西伊斯曼（Tennessee-Eastman，TE）生产流程和正在实际生产运行的的浮式储油卸油生产系统（Floating Production Storage Offloading，FPSO）典型生产流程进行了分析。利用PID控制算法对TE生产流程建立了基于MATLAB的仿真。同时，对FPSO典型生产流程进行了仿真，最后获得了两个流程在正常生产状态和故障状态下的样本数据。将主元分析和核主元分析故障检测法在TE生产流程进行了算法的验证，然后在FPSO典型生产流程进行了故障诊断的应用。通过对主元故障检测法和核主元故障检测法检测效果和故障漏检率的对比分析，得出在具有强非线性环节的生产流程中核主元故障诊断方法具有一定的优势。最后本文分析了决策导向无环图（Decision Directed Acyclic Graph，DDAG）故障辨识方法，然后将其应用到TE生产流程和FPSO典型生产过程进行了故障的辨识。结果表明其DDAG故障辨识法可以满足两个流程故障辨识正确率和辨识时间的要求。
[Abstract]:The fault diagnosis of the production process has been one of the key problems in the process industry all the time . The research and application of the fault diagnosis of the industrial process has a great promotion effect on ensuring the operation safety and improving the economic benefit . The data - driven fault diagnosis research is an important point of attention in the process industry . This paper analyzes the methods of fault diagnosis and the development trend of fault diagnosis . Through the analysis of principal component analysis ( PCA ) fault diagnosis method and Hotelling T2 statistic and SPE statistics , the fault diagnosis process is given . At the same time , the characteristics of several common kernel functions and the principle of kernel principal component analysis ( KPCA ) fault detection algorithm and its application process are analyzed . This paper analyzes the typical production process flow of Tennessee Himan ( TE ) and the typical production process of FPSO . Based on the PID control algorithm , a simulation based on MATLAB is established . At the same time , the typical production flow of FPSO is simulated , and the sample data of two processes in normal production status and fault condition is obtained . The main element analysis and kernel principal element analysis fault detection method is validated in the TE production process , and then the application of fault diagnosis is carried out in the typical production process of FPSO . Through the comparative analysis of the main element fault detection method and the nuclear main element fault detection method and the failure detection rate , it is concluded that the method of the nuclear main element fault diagnosis in the production flow with strong nonlinear link has certain advantages . In the end , the decision - oriented acyclic graph ( DDAG ) fault identification method is analyzed , then it is applied to the fault identification of TE production process and FPSO typical production process . The results show that the DDAG fault identification method can meet the requirements of the accuracy rate and the identification time of two process fault identification .

【学位授予单位】：天津理工大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TH165.3

【参考文献】