Design Of Working Condition Analysis System For Crude Distillation Process Based On Data-driven

Crude oil distillation is the first process of production in refining plants,and also the most important and key process in refinery enterprises.The stability of its operation condition has a great impact on the subsequent process.Based on the actual operation data of atmospheric and vacuum distillation process from a large refining enterprise in China,this paper studies data-driven methods,which are used for the condition analysis and fault early warning of the crude distillation process.This study can improve the operation stability and safety of the crude oil distillation unit(CDU),push the intelligent manufacturing construction of process enterprises in China.In the first chapter,the background,significance and current situation of this research are summarized,and the main problems existing in the study of operating conditions of crude distillation process are analyzed.Firstly,this chapter introduces the research background and crude oil distillation process technology,and then describes the current research status of the steady-state condition discrimination and abnormal operating condition detection at home and abroad.Secondly,it introduces the key problems existing in crude oil distillation process condition analysis.The main structure and content arrangement of this paper are given in the end.In the second chapter,the method to distinguish the operation conditions of crude distillation process is studied.Firstly,it introduces the principle and implementation steps of using principal component analysis(PCA)to distinguish the steady-state.It judges whether the operation condition is stable on the basis of synthesizing multiple variable information.Besides,it can classify and model new operation condition,so as to realize the online update and expansion of the model set of historical operation condition.Then numerical simulation is carried out to verify the effectiveness of the method.Finally,it is applied to the condition discrimination and classification of atmospheric tower in a refinery,which improves the accuracy and real-time of the steady-state condition identification and classification.In the third chapter,the method of fault early warning in crude distillation process is studied.Dynamic principal component analysis(DPCA)method is introduced at first,and it is applied to the fault early warning of the primary distillation tower.Then according to the differences of the delay time and time constant of each variable in the crude distillation process,the extended matrix composition method of DPCA is improved.Numerical simulation shows that the improved algorithm can reduce the missed alarm rate of traditional DPCA.Finally,aiming at dealing with a certain kind of sensor failure in crude distillation,the algorithm of fault data reconstruction based on DPCA is studied and verified successfully.In the fourth chapter,the fault early warning algorithm which can reduce the false alarm rate is studied.Firstly,the closed-loop control unit in CDU is analyzed,and a fault early warning algorithm which separates manipulated variables(MV)and controlled variables(CV)is proposed.Then the closed-loop control model is built to generate multiple groups of MV and CV,and numerical simulation of this method is carried out.Finally,this method is verified in failure of the primary distillation tower in a refinery.In the fifth chapter,a software platform of crude distillation process condition analysis is designed and developed.First of all,considering the production requirements of enterprises and key technology points,the overall design of the software platform is described,including database-based data storage and query,steady-state condition discrimination,fault early warning and other functional modules.Then this chapter describes the function and development process of the platform,and finally gives the actual operational effect.The sixth chapter is the summary and prospect of the whole paper.This chapter summarizes the research work,the existing problems and shortcomings,and the future research directions of this subject are pointed out in the end.The condition analysis technology of crude distillation process developed in this paper has passed experimental test in the provincial engineering technology research center jointly built with enterprises.It surely has promising industrial application value for the condition analysis of crude distillation process in the future.