Based On Semi-quantitative SDG Model Chemical Process Fault Diagnosis

Chemical industry is one of the important pillar industries in our country. It relates tovarious aspects of the economic, defense, resources and basic necessities in people’s dailylives. In the one hand, with the development of science and technology and people’s livingstandard improving, The chemical products demand are increasing in various sectors. So lotsof chemical companies expand the production scale to meet the social needs. At the same timethe processes are becoming more complex.In recent years, with the innovation of technology,especially in computer and process control, the automation level is rising in chemical industry.On the other hand, chemical industry is a high-risk industry. Chemical accidents would causegreat losses. Therefore, chemical security has been more and more attention. And it hasbecome a focus to develop the efficient real-time chemical process fault diagnosis technology.At first, this paper introduces the research background and significance of the chemicalprocess fault diagnosis. There is a brief introduction about SDG (signed directed graph) faultdiagnosis and it’s characteristics in chemical process. The domestic and international researchprogress and current status about SDG fault diagnosis are summarized.Digraph is a graph with directed arcs between the nodes and SDG is a graph in which thedirected arcs have a positive or negative sign attached to them. There are three signs aboutevery note:“－,0,＋”. Respectively, the signs represet low, nomal, high state. Secondly, inthis paper the semi-quantitative model combines SDG model with membership infuzzy theory because the traditional SDG model for fault diagnosisdoes not consider the quantitative information but rely on the causal relationship betweenvariables. The node membership means the deviation degree from the normal value in thesemi-quantitative SDG model. So the fault source can be determined effectively.The hybrid reasoning combimed positive reasoning with negative reasoning is applied in thispaper to remove the false solutions and improve the diagnostic accuracy. In other words, wefirstly apply reverse reasoning to collect the possible sources according to the known data. Wecalculate each note’s membership and use forward verification from special fault source noteto deduce the notes’ state along compatible pathways in the SDG. Then speculated each notesigns are compared with the real signs.To achieve online diagnosis, we will apply a real-time expert system to improve thespeed of semi-quantitative SDG diagnosis in this paper. We turn every SDG diagnosisconclusion to an expert knowledge rule stored in the knowledge base using the If-Then form. Finally, the application study of TE proeess verifies the feasibility and superiority of theproposed method, and achieves the expected results.