利用神经网络优化铁基软磁材料热处理工艺研究

Application of back-propagation neural network (BPNN) to the optimization of heat-treatment of Fe-based soft magnetic materials

采用熔体快淬法制备软磁非晶薄带,经不同温度和时间对材料进行热处理,获得对应不同热处理温度和时间的磁性能数据。利用基于MatLab软件平台的神经网络,将传统热处理工艺与其相结合,对铁基软磁材料热处理工艺进行优化。研究结果表明,反向传播神经网络(BPNN)能够较好地预测这种材料磁特性随热处理条件变化的规律,可用于优化铁基软磁材料热处理工艺。

Fe-based soft magnetic amorphous alloy ribbons were prepared by melt-spinning, followed by heat-treatment at different temperature for different times. Using back-propagation neural network （BPNN） based on MatLab the heat treatment technology of the alloy was optimized. The results indicated that BPNN model can successfully predict the dependence of magnetic parameters on heat-treatment conditions. BPNN is proved to be used in the optimization of heat treatment process of Fe-based soft magnetic material.