利用人工神经网络进行激光等离子体诊断  

Laser Plasma Spectral Diagnostic Using a Neural Network

作　　者：宋向阳[1] 韩申生[1] 徐至展[1] 黎忠[1] 张令清[1] 张正泉[1] 

机构地区：[1]中国科学院上海光学精密机械研究所,上海201800

出　　处：《光学学报》1996年第4期500-503,共4页Acta Optica Sinica

摘　　要：利用前馈神经网络对激光等离子体打靶实验中所得的X光光谱数据进行处理，可以方便地求出等离子体的电子温度和电子密度等参数。在对网络的训练时采用误差信号反向传输算法，训练后的神经网络能够有效地对X光光谱数据进行处理。文中给出了用此法算出的Mg等离子体电子温度和电子密度的空间分布轮廓，与用传统方法所得的结果完全吻合。A feed forward neural network is applied to the analysis of K-shell X-ray line spectra from a laser produced plasma, and to the accurate determination of the plasma parameters. After training on error backpropagation algorithm, the network can effectively produce the electron temperatures and the electron densities. The use of this method for X-ray is illustrated. Spectroscopy of a high density and high temperature Mg plasma is illustrated. The both parameters are produced, and agree well with the results from conventional method.

关 键 词：神经网络 X射线光谱 激光等离子体 诊断 

分 类 号：TL653[核科学技术—核技术及应用]