Research On Classification Algorithm Of Foodborne Bacteria In Raman Spectrometer

The detection of food-borne pathogens is an important task.Traditional laboratory testing methods can cause irreversible damage to food-borne pathogens,and the detection time is long,which is prone to false positive or false negative phenomena.Raman spectroscopy,as a kind of spectral analysis technology,has developed rapidly in recent years.Because the formation of Raman spectroscopy is related to the nature of specific molecules within the substance,different types of food-borne pathogens such as purines,peptides,proteins and other molecules The composition and nature are also different,so Raman spectroscopy can be used to detect and identify foodborne pathogens.Compared with traditional biological detection methods,the biggest advantage of Raman spectroscopy is that it can achieve non-destructive testing of samples,and the combination of Raman spectroscopy and computer systems can reduce analysis time,improve recognition accuracy,and maximize extraction and extraction.Mining effective information in the spectrum.In this paper,Raman spectra of two food-borne pathogenic bacteria,Escherichia coli O₁₅₇:H₇ and Brucella Strain 2,were collected by Changchun Veterinary Research Institute of Chinese Academy of Military Sciences.The original Raman spectrum has problems such as noise,fluorescent background,and large data dimensions.If only relying on manual spectrum identification,it will cause identification errors.In response to this problem,an intelligent Raman spectrum identification model of food-borne pathogens is established in this paper.First,the initial Raman spectrum is normalized,noise reduced,and the baseline is removed,and then feature extraction is performed using principal component analysis With dimensionality reduction,we finally studied a number of basic classifiers,inspired by artificial neural networks,and proposed a fusion model for classification recognition.The experimental results prove that the fusion model has a good classification effect,the main contents of this article are as follows:?1?The principle of Raman spectroscopy is studied,and the Raman spectroscopy is explained from electromagnetism and quantum mechanics.The working mode and parameter selection of the Raman spectrometer and its internal functional units are introduced.?2?In the preprocessing and feature extraction stages of Raman spectroscopy,the moving average filter,moving median filter,Savitzky-Golay filter are used to reduce the noise of the experimental data.The two indicators of signal-to-noise ratio and mean square error are used to select the most effective noise reduction Hershey filter parameters;using the principal component analysis method to extract features from the original Raman spectrum,the experimental results show that when the original Raman spectrum data is reduced to three dimensions,its contribution rate reaches 95.41%,which has a good spatial distinction degree.?3?KNN,logistic regression and support vector machine are used to build an intelligent food-borne pathogenic Raman spectrum recognition model.Inspired by the artificial neural network multilayer neuron structure,a fusion classification model is proposed,which uses KNN and support vector machine.As the first-level classifier,logistic regression as the second-level classifier finally outputs the classification results.The experimental results confirm that its accuracy is better than that of a single classifier,and it can accurately identify Escherichia coli O₁₅₇:H₇ and Brucella Strain 2.