Application Of The Near Infrared Spectroscopy Technology In The Rapid Determination Of Fermentation Powder

Near Infrared Spectroscopy technology,due to its rapid,non-destructive,without complicated sample processing,without the use of chemical agents,environmental pollution,the specific advantages of real-time analysis,after years of development,has become a modern analytical techniques with much attention,however there are some weaknesses with NIR technology,such as weak signal strength,very wide wavelength band and the overlap between the peaks,the serious interference situation,and therefore can not be directly related to qualitative/quantitative analysis.In this paper,we used near infrared spectroscopy technique for fermented products in a variety of levels of active ingredients modeling analysis to establish a new type of detection technology that can be widely used in the quality inspection of pharmaceutical raw materials and other fermented products industry.We used NIR spectroscopy to establish the determination of the Cordyceps acid,polysaccharides,adenosine and protein concentrations of Paecilomyces hepiali and Marasmius androsaceus mycelium quantitative analysis calibration model.The fermented products prepared with different conditions,has passed MCPLS method to identify abnormal samples,screened Savitzky-Golay Smoothing,Fast Fourier Transform,a first derivative,second derivative spectra and other pretreatment and pretreatment window size,preferred movable windows characteristic wavelength variable and other related variables by MWPLS and MW RBFNN.Da was the indicator of the two models.By comparing,the most the most suitable calibration models of Cordyceps acid,polysaccharides,adenosine and protein content of fermentation products were obtained.For determination of Cordyceps acid and polysaccharide in Paecilomyces hepialid mycelium,RBFNN model is better than others.For determination of adenosine and protein concentration,PLS model is better than others.As shown,Rc~2 values were 0.9714,0.9803,0.9761 and 0.9553 respectively,and Rp~2 values were 0.9663,0.9850,0.9728 and 0.9393 respectively.For detecting the levels of Cordyceps acid,polysaccharide and protein in Marasmius androsaceus mycelium,RBFNN models were better than others;meanwhile,PLS model was better for adenosine detection.As shown,Rc~2 values were 0.9274,0.9440,0.9399 and 0.9273 respectively,and Rp~2 values were 0.9009,0.9492,0.9437 and 0.9404,respectively.