Qualitative analysis of textile fibers based on Raman spectroscopy data augmentation
The identification of components in waste textiles has the problems of high labor cost and low efficiency.A machine learning rapid identification method based on Raman spectroscopy is a potential solution to solve this problem.However,machine learning methods typically require a large amount of data for training.In order to reduce the cost of data collection and improve the classification accuracy of models for textile fibers in small sample Raman spectroscopy datasets,a Raman spectroscopy data augmentation method was proposed.This method enhances the data on preprocessed Raman spectroscopy datasets of 5 textile fibers by specifying Pearson correlation coefficients combined with signal-to-noise ratio formula for noise superposition and linear combination of Dirichlet distribution.The results show that after data augmentation,the SVMpoly model achieves an average accuracy of 92.4%in 10 rounds of 2-fold cross validation,which is 68.2%higher than the original Raman spectroscopy dataset.This data augmentation method can enrich the diversity of samples while expanding the dataset,thereby improving the classification performance of the model.
NETL
NSTL
万方数据
