针对目前光谱特征峰识别定位技术识别率低、定位误差大、无法获得光谱线型函数等问题,提出了一种基于改进正弦余弦算法的光谱特征峰识别定位方法。该方法利用动态转换概率改进正弦余弦算法,将改进的正弦余弦算法与多种光谱线型(高斯、洛伦兹和Voigt)拟合方法相结合,通过迭代寻优计算,最终得到对应的光谱特征峰位置。该方法不仅能得到光谱特征峰的位置,而且可以得到光谱线型函数。实验证明,无论强峰、弱峰还是重叠峰,提出的基于改进正弦余弦算法的光谱特征峰识别定位方法在识别率、定位准确率、谱值拟合效果和对噪声的抑制能力等方面均有显著提高。

In order to solve the problems of traditional characteristic spectral peak recognition and positioning methods, such as low recognition rate, large positioning error, and being unable to obtain spectral line-shape function, we propose a characteristic spectral peak recognition and positioning method based on an improved sine cosine algorithm. A sine cosine algorithm is improved by dynamic conversion probability, and then combines with the fitting methods of various spectral line-type functions (Gaussian, Lorentzian, and Voigt). The corresponding positions of characteristic spectral peaks can be obtained by iterative optimization. The improved method can not only locate characteristic peaks precisely, but can also obtain the line-type function of the spectrum. The experiments show that the proposed method significantly improves the recognition rate, positioning accuracy, peak fitting effect, and noise suppression ability for strong, weak, and overlapping peaks.