Fault Diagnosis Based On Bayesian Network And Knowledge Graph In The Chemical Process

The chemical industry is an important pillar industry of Chinese economy,and its safe production is of great significance to the stable development of the economy and society.The fault diagnosis of the chemical process is one of the key links to ensure the safety of the chemical industry,and the accurate fault location and timely corresponding measures can greatly reduce the risk and benefit loss of the production process.This thesis focuses on the fault diagnosis problems in the chemical production process,including fault monitoring,fault alarm,fault root variable traceability and fault maintenance,and proposes a realization method from production data to maintenance plan.The main tasks are as follows:(1)The data collected in the chemical production process is generally of high dimensionality,strong coupling,and lack of fault tags.The conventional methods are difficult to make accurate judgments on the working states.This thesis proposes a Bayesian network-based chemical process fault monitoring scheme,which establishes a Bayesian network model of the production process,monitors the working states of each variable,and completes the root cause analysis of the fault.The TE process data is used to validate the reasoning effect and the root cause fault variable traceability results of the proposed model.It is proved that the scheme can not only realize fault monitoring,but also trigger a fault alarm in time and provide accurate fault location,which provides a basis for rapid troubleshooting of faults in the chemical process.(2)In the practical chemical production,the formulation of fault maintenance programs mostly relies on the production experience of engineers or technical manuals,which has the disadvantages of poor adaptive ability and lack of pertinence.To solve those limitations,this thesis proposed fault diagnosis method based on knowledge graph,which realizes the connection between fault monitoring results and fault knowledge,integrates various chemical production-related content,constructs a chemical process fault knowledge graph,establishes a fault information query route,and realizes the automatic formulation of fault maintenance plans.Taking the fault conditions of the TE process as an example,the maintenance plan obtained from the query result is consistent with the actual situation,indicating that the chemical process fault knowledge graph can automatically generate an accurate maintenance plan,and improve the speed and reliability of fault elimination.(3)Aiming at the problems of knowledge graph editing,such as multiple professional restrictions,high operating difficulty and fixed operating platform,this thesis designs and develops a chemical process fault knowledge management system,which realizes the functions of system management,the chemical process fault knowledge management and user management.The various functions are tested to verify the practicability of the system,which proves that the system can improve the ease of use of the chemical process fault knowledge graph,lower the threshold of using the chemical process fault knowledge graph,and provide technical support for the practical application of the chemical process fault knowledge graph.