Research On Key Technologies Of Digital Twin For Condition Monitoring And Diagnosis Of Electromechanical Systems

As high-tech-intensive artifacts,modern electromechanical products are usually large complex systems with complex functions and highly integrated multi-domains.Due to the potentially severe consequences of abnormal system states,condition monitoring and diagnostics have been introduced to ensure the reliable operation of complex electromechanical systems.Traditional model-based or data-driven methods rely on the accuracy of knowledge models and the integrity of data samples,and are susceptible to interference from actual operating conditions.As a real-time and dynamic digital description of the physical entity,the digital twin can maintain consistency with the physical entity in the virtual space and improve the data integrity by simulating various operating conditions through simulation technology.However,there are problems such as multi-disciplinary model integration,cross-phase information transfer,and insufficient fault simulation data in the application of digital twins in realizing the condition monitoring and diagnosis of electromechanical systems.To solve these problems,this thesis focuses on the key technologies of the digital twin for condition monitoring and diagnosis of electromechanical systems and conducts the following research work:(1)A digital twin architecture for condition monitoring and diagnosis of electromechanical systems is studied.Combined with the idea of systems engineering,a hierarchical digital twin paradigm is proposed,which realizes digital twin-driven design and operation collaboration.And a digital twin architecture for electromechanical system condition monitoring and diagnosis is designed,which acknowledges the integration of model and entity,normal and fault,virtual and real,laying the framework for subsequent digital twin research.(2)The construction and update methods of multi-domain models of electromechanical systems are studied.The hierarchical modeling method based on Modelica is proposed,and the multi-domain model of the electromechanical system with high fidelity is constructed under the unified model framework.The model update and parameter adjustment mechanism are designed,and the model is used as the carrier to realize the transmission of cross-stage information,which ensures the consistency of the model.Finally,an accurate model is provided for the digital twin system for the condition monitoring and diagnosis of electromechanical systems.(3)The fault behavior model derivation method based on the reuse mechanism is studied.The correlation mapping between fault information and behavior models and the automatic derivation method of the fault behavior model based on the reuse mechanism at different system levels are studied,which realizes the continuous expansion of fault information in the digital twin,and a complete and sufficient fault simulation data set is established.Finally,a solid data foundation is provided for the digital twin applications for the condition monitoring and diagnosis of electromechanical systems.(4)A digital twin fusion method for condition monitoring and diagnosis is studied.The virtual-real interaction and collaboration mechanism based on data fusion is studied to ensure the real-time consistency between the virtual models and the physical entities.A digital twin method integrating virtual and physical entities is studied,which overcomes the shortcomings of poor model accuracy and incomplete data in traditional methods,and realizes comprehensive and intuitive operation condition monitoring and more accurate fault diagnosis.(5)According to the actual research and development requirements of a certain model of civil aircraft wheel brake system,the proposed method is used to build a digital twin experimental platform,and all-digital virtual simulation and digital twin experiments are conducted.The experimental results show that the proposed digital twin method can achieve more accurate condition monitoring and diagnosis,and its accuracy reaches91.67%,which is better than other forms of the diagnosis process,verifying the effectiveness and feasibility of the proposed method in this thesis.