Fault Diagnosis And Prediction Technology Of TEG Dehydration Unit

The triethylene glycol(TEG)dehydration device is an important petrochemical production equipment,which has the characteristics of multiple equipment and complex failure mechanism.Due to its high temperature and pressure production environment,when it fails,it will cause huge economic losses and even major accidents.Therefore,in order to avoid the occurrence of major accidents and unnecessary economic losses,and to ensure its safe and stable operation,it is of great significance to conduct state monitoring and fault diagnosis of the dehydration system.At the same time,in the operation of the TEG dehydration unit in produced a large number of monitoring data,and the industrial big data contains the status information of the equipment,based on this,taking teg dehydration unit as the research object,in order to realize the teg dehydration unit of intelligent monitoring and diagnosis as the goal,to carry out the teg dehydration unit anomaly recognition and parameter prediction technology research,design and develop three glycol dehydration unit of intelligent monitoring and diagnosis system.Specific research contents are as follows:(1)Based on the technological process of the TEG dehydration unit,the typical failures of the TEG dehydration unit are analyzed.In view of the possible deviation of empirical knowledge,a steady-state and dynamic model of the TEG dehydration unit HYSYS is established,and the mapping relationship between typical failures of the dehydration unit and the monitoring parameters is improved to provide a basis for abnormal identification and fault diagnosis.(2)Aiming at data problems such as missing data and noise in the data monitoring of the TEG dehydration device,a data preprocessing method combining random interpolation method,interquartile range(IQR)criterion method and wavelet packet noise reduction is established to Improve the quality of data.In view of the dynamic change of the parameters of the teg dehydration unit of operation characteristics,put forward the principal component analysis(PCA)and symbol directed graph(SDG)anomaly identification technology,through the PCA anomaly recognition and SDG fault location of abnormal joint diagnosis for dewatering device identification,using historical failure data to verify this PCA-SDG fault diagnosis methods,the results show that PCA-SDG fault diagnosis method can effectively identify the fault.Based on the idea of monitoring the state of the intelligent monitoring system and saving case data,a threshold-based and case-based identification method is proposed.The threshold-based method identifies simple process faults by setting thresholds.The case library-based identification method compares and analyzes the abnormal data identified by the PCA with the failure case data to directly and accurately locate the fault.(3)In view of the lack of monitoring parameter state prediction in TEG dehydration unit,a vector auto-regressive(VAR)model is proposed to predict the monitoring parameters.The experimental results show that the VAR prediction model has good prediction effect and effectively predicts the change trend of parameters.Aiming at the high cost of detecting key dehydration indicators such as natural gas water dew point in the actual production process,an attentive neural processes(ANP)online prediction method driven by process monitoring parameter fusion is proposed.Firstly,the key feature parameters are selected through the gradient boosting decision tree(GBDT),and then the functional regression relationship between the key parameters and the natural gas water dew point is established adaptively through the ANP model,and the natural gas water dew point is predicted.Using historical data for experimental verification and comparison with GBDT,the results show that the ANP model after feature selection has strong generalization ability and prediction accuracy.(4)Design and develop an intelligent monitoring and diagnosis system for natural gas dehydration equipment.In order to improve the scalability of the system,a system development framework based on Spring boot is designed.At the same time,a B/S three-tier application architecture and a service-oriented architecture(SOA)development model are designed.Design the overall system architecture of the data source layer,data service layer,application service layer and interface display layer,and design the system database,Python data analysis service module and system front-end functional module respectively.Through the analysis of test cases when the intelligent monitoring and diagnosis system is online and running,the intelligent monitoring and diagnosis system of the TEG dehydration device developed in this paper meets the needs of monitoring and diagnosis of the TEG dehydration device.