Non-destructive Classification of Psychotropic Drugs Based on Spectroscopy from the Perspective of Public Security

KeyWord:spectral fusion  public security  psychotropic drugs  Bayes discriminant analysis  multilayer perceptron

Abstract:

      In order to maintain the public safety and order of the society£¬and achieve fast non-destructive and accurate qualitative analysis for drugs£¬the infrared spectrum data for benzodiazepines and phenothiazines samples were collected in the experiment to construct different data classification models and compare the effects of different spectral data sets on the accuracy of their analysis£®In the process of identifying and classifying the two types of psychotropic substances£¬benzodiazepines and phenothiazines£¬the second-derivative model based on the full spectra was better than other single models£¬whose overall classification accuracy rate reached 92.7%£®Combined with Bayes discriminant analysis£¬the second-derivative classification model based on the full spectra and fingerprint spectra fusion performed the best in the process of identifying and classifying the two types of psychotropic substances£¬benzodiazepines and phenothiazines£¬with an overall classification accuracy rate reaching to 100%£®Meanwhile£¬in the process of identifying and classifying different kinds of psychotropic substances in the same type£¬its overall classification accuracy rates for various kinds of psychotropic substances in benzodiazepines and phenothiazines reached 96.7% and 100%£¬respectively£®This research realized the rapid and accurate qualitative analysis of different types£¬as well as different kinds of sleeping pills of the same type£¬providing a certain technical support for In order to maintain the public safety and order of the society£¬and achieve fast non-destructive and accurate qualitative analysis for drugs£¬the infrared spectrum data for benzodiazepines and phenothiazines samples were collected in the experiment to construct different data classification models and compare the effects of different spectral data sets on the accuracy of their analysis£®In the process of identifying and classifying the two types of psychotropic substances£¬benzodiazepines and phenothiazines£¬the second-derivative model based on the full spectra was better than other single models£¬whose overall classification accuracy rate reached 92.7%£®Combined with Bayes discriminant analysis£¬the second-derivative classification model based on the full spectra and fingerprint spectra fusion performed the best in the process of identifying and classifying the two types of psychotropic substances£¬benzodiazepines and phenothiazines£¬with an overall classification accuracy rate reaching to 100%£®Meanwhile£¬in the process of identifying and classifying different kinds of psychotropic substances in the same type£¬its overall classification accuracy rates for various kinds of psychotropic substances in benzodiazepines and phenothiazines reached 96.7% and 100%£¬respectively£®This research realized the rapid and accurate qualitative analysis of different types£¬as well as different kinds of sleeping pills of the same type£¬providing a certain technical support for the accurate qualitative analysis of such smuggling controlled drug cases£®