Research And Application Of Fault Diagnosis Method For Centrifugal Pump Impeller Based On Digital Twinning Technology

Centrifugal pump equipment is an indispensable liquid transportation and pressurization equipment in modern industrial production and manufacturing processes.Its health status directly affects the production efficiency and economic benefits of enterprises,and the importance of modern industrial production processes is self-evident.However,the current domestic technical means for fault diagnosis,health management,and maintenance of complex mechanical structures within centrifugal pumps during their full life cycle evolution and development are not at a high level of digitization and intelligence,making it difficult to meet the new requirements of modern industrial development in the context of new technologies.In recent years,the rapid development of technologies such as digital twins and artificial intelligence has undoubtedly injected new impetus into this.Turbomachinery is a core component that bears the comprehensive effects of centrifugal force,fluid dynamics,vibration,thermal stress,and media,and is most prone to abnormalities and failures.Therefore,this article takes the centrifugal pump in the aviation coal pipeline transportation of a petrochemical enterprise as the research object,with the aid of digital twinning technology,and around the construction of the centrifugal pump digital twinning system,the digital realization of turbomachinery fault diagnosis,etc.,a relatively systematic study has been conducted.The main work includes:1)Aiming at how to carry out fault diagnosis of centrifugal pump turbomachinery based on digital twin technology.This article first provides a detailed overview and analysis of the specific content of each dimension of the digital twin five-dimensional model,and extends its standard technical architecture to the research of centrifugal pump digital twin technology.Then,specific application analysis was conducted from the five dimensions of centrifugal pump physical equipment,digital twin model,twin data,application services,and data connectivity interaction,forming the main research line and specific digital twin architecture of this article.Finally,through analysis and demonstration of relevant modeling methods,simulation technologies,and system construction platforms,the implementation scheme of the core content of the twin architecture is determined,laying a foundation for subsequent research work.2)Aiming at the modeling of the digital twinning system of centrifugal pump,on the one hand,in order to ensure the appearance twinning of the physical entity of centrifugal pump,under the premise of fully considering various objective factors,the point cloud reverse modeling technology is adopted to establish the highly fidelity appearance model of centrifugal pump.On the other hand,in order to establish a twin attribute model that can explain and describe the hydraulic characteristics,physical phenomena and other flow properties of the internal complex mechanical structure flow field of centrifugal pump during operation,a digital twin attribute modeling method of centrifugal pump based on CFD and data-driven compensation is proposed.CFD numerical simulation ensures the interpretability of the physical properties of the internal flow field of centrifugal pump,The data-driven error compensation model further improves the prediction accuracy of hydraulic characteristics of outlet pressure.3)Aiming at the fault diagnosis of centrifugal pump impeller machinery,a fault diagnosis method based on Yolov 5 target detection was proposed according to the data characteristics of the digital twin model output nephogram.Firstly,a digital twin model for centrifugal pumps is used to simulate the random fracture failure of impeller blades and generate pressure and velocity nephogram of impeller flow field with different fault characteristics;Secondly,based on the learning and training of the Yolov 5 algorithm,two types of machine vision models,pressure and velocity nephogram,were obtained,and combined with statistical analysis to achieve preliminary diagnosis of turbomachinery faults;Thirdly,considering the complementary advantages of the two types of detection models,the results of the two models are fused based on the idea of stack integration to improve the accuracy of impeller fault diagnosis.Finally,through experimental simulation analysis,the Yolov 5 target detection method can effectively detect abnormal blades and preliminarily diagnose impeller faults more accurately,while stack integration further improves the accuracy of fault diagnosis.4)In order to apply the digital twin model of centrifugal pump and the fault diagnosis method of turbomachinery based on the model to the engineering practice,combined with the actual needs of the project and the system feasibility analysis,based on the research work of 2)and 3),and with the help of technologies and software platforms such as Ansys,My SQL,Unity,the system landing,centrifugal pump operation monitoring,prediction guidance The digital twin system of centrifugal pump with the functions of fault diagnosis and feedback can help users understand the operation status of centrifugal pump system and the occurrence of abnormal impeller machinery through the intuitive graphical and visual interface,thus realizing the digitalization and intelligence of the health maintenance and management process of centrifugal pump equipment.