非线性数据拟合对重叠峰信号的分离  

Separation of Overlapping Peak Signals by Non-linear Fit

作　　者：袁红志[1] 游开明[1] 谭延亮[1] 

机构地区：[1]衡阳师范学院,湖南衡阳421008

出　　处：《衡阳师范学院学报》2013年第3期37-39,共3页Journal of Hengyang Normal University

摘　　要：从最小二乘法出发提出了一种含噪音重迭峰信号的分离方法-利用非线性数据拟合处理含噪音的重叠峰数据,得到各子峰的函数表达式,将传统的峰分辨和峰分解两个步骤合二为一,简化了重叠峰信号分析过程。讨论了该方法对不同分离度重叠峰信号分离的效果,通过仿真分析发现该方法在对分离度仅为0.75的重叠峰信号的分离也有很好的效果。在分离度相同时,对高斯-洛仑兹峰重叠信号分离的相对误差小于对高斯-高斯峰重叠信号的分离;分离得到的各子峰参数相对误差随分离度的提高而减小。该方法可用于实际的含噪音重迭峰信号进行处理。The paper Propose a separation method for noised overlapping peak signals from feast squares method, use non-linear fit to deal with the noised overlapping peaks data. The function expression of every peak can be obtained. Thus, the two traditional process-peaks differentiation and peaks analysis is united, it predigest the analysis course of overlapping peak signals. The separation effect of this method for overlapping peak signals with different separation degree are discussed. The separation effect is well even the separation degree of the o- verlapping peak signals is 0.75. The relativity error separated overlapping peak for gaussian-lorentzian signals is less than that for gaussian- gaussian signals as the separation degree equal. The relativity error of every peak parameters separated overlapping peak signals will be less while the separation degree grows. This method can be used to deal with the noised actual overlapping peak signals.

关 键 词：非线性拟合 最小二乘法 重叠峰信号分离 

分 类 号：O422.7[理学—声学]