Study On Pollen Classification Using Dual Wavelength Excitation Fluorescence Spectroscopy

With the improvement of the living standard,people pay more and more attention to their health.Because of the high potential allergenicity of pollen,there is a great threat to people's work and life quality.Therefore,it is of great practical significance and value to research and develop the technology to detect the pollen particles in the air.Based on the principle of intrinsic fluorescence,a set of dual wavelength excitation pollen detection device was developed to measure the fluorescence spectrum of pollen.Our results provide a great help for the detection of allergic pollen in the atmosphere based on fluorescence detection.Using fluorescence spectrophotometer to measure the excitation-emission matrix of the 6 kinds of pollen,and through the excitation-emission matrix and the total fluorescence intensity of 6 kinds of pollen,we found the optimal excitationwavelength and emission wavelength range required for the identification and classification of pollen analysis.Two different excitation wavelengths in the optimal excitation wavelength range were selected.An unsupervised pattern recognition analysis,i.e.,principal components analysis,was used at the pollen fluorescence spectra with different solvents under the two excitation wavelengths,respectively.It is qualitative proved the feasibility of the dual wavelength excitation spectrum analysis for the pollen classification.Based on LED and LD light sources,a miniaturized and low cost fluorescence detection system was established.Firstly,the data of pollen fluorescence spectroscopy were measured and the principal component analysis was used to qualitatively view the classification results,and then a partial least squares discriminant analysis was used to analyze the modeling results and predicting results of experimental data for quantitative analysis of supervision in pattern recognition.The final success rate of modeling of the 6 kinds of pollen was 100%,while in the cross validation,the success rate of predicting of the other 5 kinds of pollen was up to 100%,except for 1 kinds of pollen,which is 96.67%.