Study On The Fast Evaluation Method Of AC Loss Of High Temperature Superconducting Magnets

AC loss is the main heat source during the dynamic operation of superconducting magnets.Its size and distribution directly affect the thermal stability of high temperature superconducting magnets.Rapid and accurate evaluation of AC losses during dynamic operation of superconducting magnets can help to obtain the current range and change rate of safe operation of superconducting magnets,optimize the structure of superconducting magnets and parameters of cryogenic system,reduce thermal losses during dynamic operation,improve thermal stability of superconducting magnets and reduce the pressure of quench protection devices.It is necessary to monitor the AC loss of superconducting magnets in real time.However,the superconducting magnets for practical engineering applications are often large in scale.The finite element method widely used in the calculation of AC loss at present has the problems of large amount of calculation and slow calculation speed.It is difficult to realize the real-time evaluation of AC loss of superconducting magnets.In this paper,a multi-scale modeling method for fast calculation of AC losses is proposed.The accuracy of the multi-scale model is improved by a series of improved background magnetic field estimation methods.On this basis,the fast evaluation of AC losses is realized by using the neural network model.The specific work is as follows:(1)The basic idea of multi-scale model and the modeling method of traditional multiscale model are introduced.The principle of the uniform current density method,which is used to estimate the background magnetic field,is summarized.The AC losses of single solenoid HTS coils are calculated by using multi-scale model.The multi-scale model,homogenization model and H equation model are compared and analyzed.(2)The idea of improving the background magnetic field estimation method of multiscale model is expounded,and a series of improved background magnetic field estimation methods are put forward.The modeling process is given.Taking the calculation results of H equation model as reference,the characteristics of different background magnetic field estimation methods in calculation speed,accuracy and application range are analyzed and compared.(3)A real-time AC loss prediction tool based on neural network model is developed.A 150 kJ high temperature superconducting magnet AC loss database under different operating conditions is constructed by using multi-scale model.The mapping relationship between AC loss and operating conditions is obtained.The AC loss under new operating conditions is predicted by using the neural network model.The accuracy and applicability of the prediction model are verified by referring to the calculation results of the homogenization model.