Research And Design Of Smart Auxiliary System For Inner Operation Monitoring In Crude Distillation Units

Crude distillation is the primary segment of refining process and its stability is significant for subsequent processes.By equipping inner operators with intelligent auxiliary monitoring system,potential risks can be detected in time.This paper studies fault detection methods based on the operating data of crude distillation units(CDU).It aims to ensure the stability of CDU and provide a solid foundation for improving the quality and efficiency of the refining enterprises.In the chapter one,the background of this thesis is introduced and the crude distillation schemes are briefly presented.The research progress of fault detection for industrial plants in recent years is also summarized.In terms of CDU,key problems of current fault detection methods are analyzed.In the second chapter,a fault monitoring method of CDU based on slow feature analysis(SFA)is studied.In this part,the Tennessee Eastman dataset,as an international standard dataset,is used to verify the effectiveness of SFA.On this basis,combined with the dynamic characteristics of process,dynamic SFA is proposed.Finally,a real flooding case of primary distillation tower is applied to prove the reliability of dynamic SFA in CDU.In the third chapter,a fault detection method for key variables in CDU is designed.Aiming at the problem that online analyzers need to be maintained frequently,a fault isolation algorithm based on SFA is studied to identify the state of online analyzers.By isolating fault variables,it provides reference for inner operators to deal with faults in time and further diagnose causes.Based on this isolation method,a fault detection algorithm for key parameters in CDU is designed.Finally,the effectiveness of this algorithm in CDU is verified by simulating a real control process of the condensation point of second line oil in atmospheric tower.In the fourth chapter,an improved SFA algorithm is put forward.Firstly,orthogonal signal correction(OSC)algorithm is analyzed and an improved SFA is proposed based on OSC.Combined with the multiblock theory,fault detection method for key parameters is designed.This algorithm only relies on historical data of key variables for modeling,without using online analyzers to get operating data.This algorithm is applied to monitor the dry point of naphtha then.It successfully verifies the advantages of this algorithm that the inspection pressure of inner operators on key parameters is further reduced.In the fifth chapter,an operating mode analysis platform of CDU is designed and developed.This platform uses JAVA,combined with MATLAB for software development.By using database,the storage and acquisition of historical or real-time data are realized.It offers a one-stop service for inner operators to master production status.Main functions such as fault alarming,patrol inspection analyzing and experience database are provided.The actual working condition of this platform is showed by presenting some webpages.In this thesis,a smart auxiliary system for inner operation monitoring in CDU is researched and developed.A fault monitoring model is established through historical data.It is used to monitor the key parameters online that inner operators need to pay attention to.This system is conducive to discover the potential faults in CDU,which is of great value for stabilizing the production of refining enterprises and improving their benefits.