基于小波变换的Stark展宽算法的优化研究  

Improvement of Stark broadening algorithm based on wavelet transform

作　　者：章志涛 丁芳 罗宇 陈夏华[1,2] 叶大为 张青 胡振华 罗广南[1] ZHANG Zhi-tao;DING Fang;LUO Yu;CHEN Xia-hua;YE Da-wei;ZHANG Qing;HU Zhen-hua;LUO Guang-nan(Institute of Plasma Physics,Chinese Academy of Sciences,Hefei 230031;University of Science and Technology of China,Hefei 230026)

机构地区：[1]中国科学院等离子体物理研究所,合肥230031 [2]中国科学技术大学,合肥230026

出　　处：《核聚变与等离子体物理》2023年第2期212-218,共7页Nuclear Fusion and Plasma Physics

基　　金：国家自然科学基金(11575243);国家重点研发计划(2017YFE0301300)。

摘　　要：为了提升Stark展宽计算等离子体电子密度的准确性,基于小波阈值去噪处理方法,对EAST氘原子光谱信号进行了处理,以信噪比(SNR)和均方根误差(RMSE)作为滤波效果的评价依据,通过对比确定了最优小波基db4,最优小波分解4层。根据噪声估计值计算适合的阈值参数进行信号重构,并将经过去噪处理后的数据应用到后期Stark展宽算法计算等离子体密度的分析过程中。结果表明,小波硬阈值去噪能够有效提高光谱信号信噪比、降低均方根误差,在消除光谱信号中噪声的同时,最大程度保留了有用的光谱细节特征信息,进而有利于光谱数据的建模效果,获得更为准确的等离子体电子密度。To improve the accuracy of the calculation of plasma electron density by Stark broadening,based on the wavelet threshold denoising processing method for EAST divertor deuterium spectral signals and using the signal-to-noise ratio(SNR) and root mean square error(RMSE) as the evaluation basis for the filtering effect,the optimal wavelet basis db4 is determined by comparison,and the optimal wavelet decomposition is 4 layers.The adaptive threshold parameter is calculated according to the noise estimation to reconstruct the signal,and the denoised data is applied to the analysis process of the plasma density calculated by the Stark broadening algorithm.The results show that the wavelet hard threshold denoising can effectively improve the signal-to-noise ratio of the spectral signal,reduce the root mean square error,and eliminate the noise in the spectral signal while retaining the useful spectral detail feature information to the greatest extent,which is beneficial to the spectral data modeling effect,and the more accurate plasma electron density is obtained.

关 键 词：小波变换 EAST 偏滤器光谱信号 Stark展宽 电子密度 

分 类 号：TN911.4[电子电信—通信与信息系统]