Rapid and accurate analysis of powder samples is crucial for public safety and food security. Raman spectroscopy, noted for its many advantages, traditionally faces issues of low sensitivity and limited detection information when analyzing powder samples. This paper explores a detection method for powder samples using ultraviolet (UV) Raman-fluorescence spectroscopy. By leveraging the principles of Raman and fluorescence spectroscopy, a combined detection optical path was designed. Utilizing a 266 nm laser as the system light source, a UV Raman-fluorescence spectroscopy detection setup was constructed and seven representative powder samples were analyzed. The experimental results demonstrate that the setup effectively detects the Raman spectral features of inorganic powders like baking soda and extracts the fluorescence spectral features of organic powders such as 1,2,4,5-tetramethyltoluene. Additionally, it accurately analyzes the Raman and fluorescence spectral features of amino acids and glyphosate preparation powders. For complex powder samples with issues such as “substances with the same spectrum” and “complex and similar spectra”, UV Raman-fluorescence spectroscopy enhances spectral information dimensions, compensating for the limitations of traditional Raman spectroscopy. Furthermore, a convolutional neural network algorithm was employed to analyze the measured data of the seven powder samples increased the average recognition accuracy (AVG) from 62.53% to 98.41%. This improvement further validates the effectiveness of the proposed method in recognizing the properties of powder samples.
Read full abstract