Application Of Dynamic Uncertain Causality Graph To Chemical Fault Diagnosis

Along with the expansion of the production scale of the modern chemical process, its complexity and risk is becoming more and more high, and the production process is in a continuous operation for a long period of time. Once a fault occurs, huge losses will be caused. As a result, the reliability and security of the chemical process has been given much attention.Dynamic Uncertain Causality Graph (DUCG) is a new methodology to deal with knowledge representation and reasoning of dynamical uncertain causalities. It is a combination of qualitative analysis and quantitative analysis and makes up for the deficiency of the pure qualitative method or quantitative method. It has many advantages.In this paper, taking the TE Process as the test platform, DUCG theory was applied to the real-time process monitoring and fault diagnosis in chemical process for the first time. The main work is as follow:First of all, this paper makes a research review in the field of fault diagnosis technology at home and abroad.Secondly, makes research on Dynamic Uncertain Causality Graph theory. DUCG is able to compactly and graphically represent uncertain causalities, simplify the graphical knowledge base based on the online evidence and expand events as independent random event expressions, and finally get the qualitative reasoning results (the set of the possible hypotheses) and the probabilities of these hypotheses.Thirdly, taking the widely used TE (Tennessee Eastman) Process as the test platform, we build the knowledge base for the new real-time process monitoring and fault diagnosis applied to chemical process based on DUCG and perform the real-time online fault diagnosis. The results show that the fault diagnosis method based on DUCG for chemical process is effective.Fourthly, in the fault diagnosis tests of the TE chemical process, the present DUCG theory fails to apply to the fault of shock type. To solve this problem, this paper proposes an improved algorithm of DUCG. The improved algorithm can be used to diagnosis the fault of shock type correctly.