Abstract: Digital twin technology is not only the core of realizing information and physical integration,but also the key to realizing digital fault diagnosis.To realize real-time mapping,real-time fault prediction and fault information feedback in time of physical space and information space,the digital twin-driven fault diagnosis method for centrifugal pump units was proposed.The digital twin technology was used to construct the digital twin mapping model of centrifugal pump unit.Based on the digital twin mapping model,the fault prediction was realized in real time by the data-driven fault diagnosis method,the fault result verification method of model simulation was used to complete the fault result verification,and the verification result was used as the condition for digital twin model correction and deep learning model adjustment.The fault diagnosis system was developed with the help of Unity3D platform,and the feasibility of the system was verified by three working conditions.

Key words: fault diagnosis, digital twin, deep learning, centrifugal pump

摘要： 数字孪生技术不仅是实现信息物理融合的核心,还是实现数字化故障诊断的关键。为了实现物理空间和信息空间的实时映射、故障实时预测、故障信息及时反馈,提出数字孪生驱动的离心泵机组故障诊断方法。首先,利用数字孪生技术构建离心泵机组数字孪生映射模型。然后,基于数字孪生映射模型,通过数据驱动的故障诊断方法实现故障实时预测,利用模型仿真的故障结果验证方法完成故障结果验证,以验证结果作为数字孪生模型修正和深度学习模型调整的条件。最后,借助Unity3D平台实现故障诊断系统的开发,并通过3种工况验证了系统的可行性。

关键词: 故障诊断, 数字孪生, 深度学习, 离心泵机组