Method and system for predicting sintered SmCo magnetic performance based on neural network

Abstract

The invention provides a method and system for predicting the sintered SmCo magnetic performance based on a neural network and relates to the technical field of magnetic materials and machine learningapplication. According to the components and technological parameters of the sintered SmCo permanent magnet, the magnetic parameters of the sintered SmCo permanent magnet are accurately predicted, the components comprise the weight percentage contents of Zr, Cu and Sm elements, and the technological parameters mainly comprise the solid solution temperature, the solid solution time, the sinteringtemperature, the secondary sintering temperature, the secondary sintering time, the pre-aging temperature, the pre-aging time and the aging temperature. And the four core performance parameters of theresidual magnetism, the coercive force, the maximum magnetic energy product and the squareness of the magnet are predicted by integrating the components and the technological parameters. On the basisof the principles of feedforward transmission and back propagation, an artificial neural network model is constructed; a sampling method of an activation function and a training set is optimized, sothat the model achieves ideal fitting and prediction effects.

Description

technical field [0001] The invention relates to the application fields of magnetic materials and machine learning technology, in particular to a method and system for predicting the magnetic properties of sintered SmCo based on a neural network. Background technique [0002] Permanent magnetic materials have the function of mutual conversion between mechanical energy and electrical energy, so using their energy conversion function and combining various physical effects of magnetism, permanent magnetic materials can be made into various permanent magnetic functional devices. Permanent magnet materials have become an important material basis for high-tech, emerging industries and social progress. The rare earth permanent magnet material prepared by rare earth and transition elements is a kind of permanent magnet material with the highest coercive force and the largest magnetic energy product. The second generation of rare earth permanent magnets is represented by sintered SmC...

AI Technical Summary

Problems solved by technology

At present, there is no report on the use of neural network models to predict the magnetic properties of sintered SmCo permanent magnets. The composition of sintered SmCo is relatively complex, and the process requirements for samples with different compositions vary greatly. Relying on experimental methods, it is limited by factors such as cycle time and cost. , it is difficult to effectively determine the optimal process conditions for specific components

[0004] In summary, the technical deficiencies of the existing background technology can be summarized as follows: (1) there is no report for the neural network model of the sintered SmCo permanent magnet

(2) The model optimization method is too simple, in order to obtain a good generalization ability, it has a great impact on the training accuracy

(3) The current neural network model for predicting permanent magnets only uses process parameters as the input of the model, without considering the joint influence of composition and process on magnetic properties

Images