基于微分进化的CMAC的束流中心轨道自动校正算法研究

Research on Electron Beam Center Trajectory Correction Algorithm Based on Differential Evolution CMAC

  • 摘要: 在高压加速器的束流引出过程中,会出现束流中心轨道偏移的现象。本文基于小脑模型神经网络(CMAC)研究束流中心轨道的自动校正算法。CMAC在学习过程中,一般采用梯度下降法更新网络权值,学习率对收敛速度影响较大。提出采用微分进化算法对CMAC的网络权值进行更新,并与传统方法进行比较。实验表明,基于微分进化的CMAC学习算法,收敛速度更快,可用于束流中心轨道的自动校正算法。

     

    Abstract: In the process of beam extraction, the electron beam trajectory is deviated sometimes in the electron accelerator. In the paper, a electron beam center trajectory correction algorithm was researched based on CMAC. In the process of learning of CMAC, it usually adapts the steepest descend method to update the weight of CMAC. The learning parameter affects the convergence rate in the CMAC. This paper presents differential evolution to update the weight of CMAC. The experiments show that the proposed method is faster than traditional method in the convergence rate. It can be used in the electron beam center trajectory correction algorithm.

     

/

返回文章
返回