Aiming at identifying and evaluating the vibration shapes of unbalanced operation windings of transformer

the characteristic information is extracted and the information-physical fusion technology is studied. Virtual/physical entities of transformer are divided into electromagnetic and mechanical subsystems

and the dynamic experimental platform is constructed to obtain multi-source and multi-feature physical information to form a global modal information base. The electrical and vibration signals are extracted as the key feature parameters of winding information fusion

and the fusion feature index is constructed to quantitatively evaluate the corresponding shapes of electromagnetic-mechanical sample space. On this basis

an Autoformer generalization model is proposed to cluster the shape domains mapped in each sample space

forming a vibration shape recognition scheme based on information-physical fusion. Finally

the validity and feasibility of information-physical fusion recognition is verified by multi-model and multi-scene virtual-reality consistency. The results show that the proposed method can effectively utilize the measurable winding electrical information to describe the unobservable mechanical feature information

more accurately reflecting the winding vibration instability shape under the unbalanced condition of the transformer

which is more advantageous than the traditional electrical parameter identification method

and lays the foundation for the digital twin of the whole life cycle of the electrical equipment.