利用多数据处理方法提高LIBS谱信号质量

发布时间：2018-03-13 15:07

本文选题：激光诱导击穿光谱　切入点：基线校正　出处：《红外与激光工程》2014年11期 　论文类型：期刊论文

【摘要】：基于分段光谱特征值提取法和小波变换算法等多个数据预处理方法,分别针对分段基线差异及光谱噪声等严重影响激光诱导击穿光谱(LIBS)信号质量的主要影响因素,开展光谱信号预处理研究。基于实验室LIBS实验装置,通过实验验证,基于多通道光谱仪不同波段光谱特征值提取,提出了一种简单易行的多组数据中特征值点连接的方法,有效地提高了LIBS光谱信号的基线平直度,并得出以小波变换算法进行LIBS谱线信号去噪的最佳算法参数。在上述工作的基础上,使用基于误差反向传播的人工神经网络方法,实现了纯铜和不锈钢等物质种类的有效识别,研究结果表明,综合利用多数据处理方法进行LIBS技术中光谱信号处理可以有效提高谱线分析和识别的质量。
[Abstract]:Based on segmented spectral eigenvalue extraction method and wavelet transform algorithm, the main influencing factors such as segmented baseline difference and spectral noise, which seriously affect the quality of laser-induced breakdown spectrum (LIBS) signal, are analyzed. The spectral signal preprocessing is studied. Based on the laboratory LIBS experimental device and the experimental verification, based on the extraction of spectral eigenvalues in different bands of multichannel spectrometer, a simple and easy method to connect eigenvalue points in multi-group data is proposed. The baseline flatness of LIBS spectral signal is improved effectively, and the optimal algorithm parameters of LIBS spectral signal denoising by wavelet transform algorithm are obtained. On the basis of the above work, an artificial neural network method based on error back propagation is used. The effective identification of pure copper and stainless steel is realized. The results show that the spectral signal processing in LIBS technology can effectively improve the quality of spectral line analysis and recognition.
【作者单位】：北京交通大学电子信息工程学院;北京交通大学理学院;
【基金】：国家自然科学基金(61177082,61205074) 北京市自然科学基金(4122063)
【分类号】：TN911.74

【共引文献】