Research On Adaptive Peak Matching Algorithm Of Vibrational Spectrum Based On Genetic Algorithm And Conjugate Gradient Descent

Raman spectroscopy is a new generation of analytical chemistry and is widely used in the fields of biology,chemistry and physics as an effective means of detecting intramolecular vibrations.The ability to accurately determine the molecular structure is called molecular fingerprinting.Raman spectroscopy,like other spectroscopic techniques,has the problem of inaccurate resolution caused by spectrum overlap when resolving complex spectra.This paper is based on this overlapping peak problem.Based on the genetic algorithm in the intelligent algorithm,combined with the conjugate gradient descent method,the peak fitting algorithm is designed based on the peak recognition of the wavelet transform,and the mutation probability in the genetic algorithm,The three parameters of crossover probability and population size were studied and compared in detail,and a set of adapted parameters was summarized for the peak fitting process of Raman spectra.The peak-fitting process with the highest efficiency was obtained when the mutation probability was 0.7 and the cross-range probability was 0.4-0.6.And for the results of peak fitting in each parameter,that is,the peak fitting quality factor was discussed,and found that regardless of the parameters,the final quality factor is constant,indicating that the algorithm designed in this paper is a good black box algorithm.