Dynamic Correlation Q-SDG Analysis Based Fault Diagnosis In Chemical Process

The fault diagnosis of chemical production process has always been a pivotal technology to ensure the safe and stable operation of chemical plants.Abnormal factors may cause turbulence in the entire system,and even cause catastrophic events.Therefore,monitoring the deterioration trend of chemical process abnormal conditions,identifying the cause of abnormal operating conditions in a timely manner,and controlling the occurrence of system failures from the source are the key to ensure the safe and normal production of chemical processes.In this paper,the correlation analysis method is used to detect and identify chemical process faults,and the potential causes of fault propagation are identified with the Signed Directed Graph(SDG)model.An intelligent identification method based on Q-SDG correlation analysis is finally proposed for chemical process.The in-depth correlation characteristics of the system are exploited to facilitate the fault memory function.The purpose of fault diagnosis and source positioning has been finally achieved.The normal operation of the chemical process must follow the laws of thermodynamics,conservation relations,etc.,leading to complex correlation characteristics among the process variables.Therefore,the failure of the chemical process is often a chain effect of information transfer between variables.From the perspective of the whole process monitoring,this paper discusses the correlation and linkages laws between variables based on similar local close neighbor structure criteria.First,the Pearson correlation coefficient is selected to initially optimize the variable selection.Then,the process feature mechanism information is extracted by using the PCA(Principal Component Analysis)contribution method.Pivotal variables with a large weight are selected from the multi-level correlation coefficient set to deduce the correlation rules of different states through the Markov transition principle analysis system.The transition status thus is identified timely.Finally,The internal information transfer process of complex systems is excavated by SDG method.The fault sources of abnormal state with compatible pathways is located accurately.And the fault memory is achieved by gather weighting coefficient Q indicators at last.The TE case application results show that dynamic correlative Q-SDG analysis based fault diagnosis method is relatively simple in modeling and diagnostic procedures.It can effectively extract process information,quickly locate fault conditions and accurately obtain fault sources,with a good process monitoring performance.