Study On Dynamic Characteristics Of Magnetic Domains Structure During Magnetization Process Of Electrical Steel Sheet At Mesoscopic Scale

Electrical steel sheet has excellent soft magnetic properties as the core material of a lot of electrical equipment such as transformers,generators and motors.Its magnetic properties also affect the operating performance of electrical equipment.The in-depth study of magnetic properties of electrical steel sheet is of great significance to improve the performance of electrical equipment and reduce losses.Magnetic domains are the mesoscopic structure inside magnetic material（mesoscopic is the scale between the microscopic and the macroscopic）.The magnetic material will be accompanied by the change of the internal magnetic domains structure and arrangement during magnetization process,and dynamic evolution characteristics of the mesoscopic magnetic domains during magnetization process can directly reflect the macroscopic magnetic properties of material.Therefore,it is of great significance to carry out the research on the dynamic characteristics of the magnetic domains structure during magnetization process of electrical steel sheets at the mesoscopic scale for the research and development of high-quality electrical steel sheets and the efficient operation of electrical equipment.In this thesis,the evolution characteristics and characterization methods of internal magnetic domains of electrical steel sheet under an external magnetic field were deeply studied.The evolution images of internal magnetic domains of electrical steel sheet under an external magnetic field were obtained by a magneto-optical Kerr microscope,and deep learning was used to predict the limited magnetic domains images of electrical steel sheet.On this basis,a method for characterizing the magnetization process of electrical steel sheet using the magnetic domains area was proposed,which laied the foundation for further research on the macroscopic hysteresis model based on the magnetic domains magnetization mechanism.The main research contents of this thesis are as follows:Firstly,the dynamic evolution process of magnetic domains inside electrical steel sheet during the magnetization process was observed and explored.After repeated polishing,samples of different types of electrical steel sheets（including grain-oriented electrical steel sheets and non-oriented electrical steel sheets）were prepared.The magnetic domains observation system based on magneto-optical Kerr effect was used to observe the dynamic evolution process of magnetic domains inside electrical steel sheet under different magnetization directions and different magnetic field intensities,and find out its evolution law.At the same time,an image information database of magnetic domains structure evolution of electrical steel sheet was established,which provided basic data for subsequent images prediction of magnetic domains structure and characterization of the magnetization process of electrical steel sheet.Secondly,based on the observed magnetic domains images,two deep learning theories were compared and studied,and the prediction of dynamic evolution images of magnetic domains structure of electrical steel sheet during magnetization process was realized.Based on the limited series of magnetic domains images of electrical steel sheets obtained experimentally,image processing technology was used to convert the images into data that can be processed by deep learning.The processed data was put into two deep learning neural networks for training,and then the trained model was used to predict the magnetic domains structure of electrical steel sheet.The comparison with measured results shows that Con LSTM can predict the dynamic evolution characteristics of the magnetic domains magnetization of electrical steel sheet more effectively.Finally,based on the above observed and predicted magnetic domains images information,the characterization method of mesoscopic magnetic domains evolution characteristics during the magnetization process of electrical steel sheets was studied.By comparing the evolution law of the width of light and dark stripes in magnetic domains images representing different magnetic moment directions,the area information of magnetic domains stripes was extracted,and the area calculation formula was given.A method to characterize the magnetization process and the state of magnetic domains of electrical steel sheets was proposed with magnetic domains area as a characteristic quantity,and the mesoscopic magnetic properties of electrical steel sheets were described.