Study And Application Of Classification Algorithm Based On Laser-induced Breakdown Spectroscopy

Laser-induced breakdown spectroscopy(LIBS)technology has the advantages of rapid,in-situ,real-time,and simultaneous detection of multiple elements,which is widely used in many fields such as industrial production,environmental monitoring,medicine and food.The detection accuracy in qualitative identification of LIBS technology will be affected by the nature of the sample,equipment and environment,which will cause spectral fluctuations.Hardware equipment and data processing have been applied to accuracy improvement.Hardware equipment modification space is small and expensive,and data processing can have low hardware costs and high recognition efficiency.Therefore,this paper analyzes the LIBS classification algorithm through theoretical analysis,and selects the MSC-IGA-SVM(Multiplicative signal correction-improved genetic algorithm-support vector machine,MSCIGA-SVM)of high classification accuracy,wide application range and strong anti-interference ability to identify the adulteration of yam decoction pieces and the non-destructive identification of citrus sugar grading.The results are as follows:(1)In the LIBS classification algorithm,the MSC-IGA-SVM model is optimized by analyzing the preprocessing,feature extraction and pattern recognition algorithms of spectral data through principle analysis.Among them,MSC can effectively reduce the volatility of various spectra without parameter optimization.IGA can retain the nonlinear relationship of characteristics while avoiding local optimal solutions.SVM has strong anti-interference ability and computational efficiency.Therefore,the preferred MSC-IGA-SVM model in theoretical analysis has higher classification accuracy,wider application range and stronger anti-interference ability.(2)The whole process algorithm of classification in this paper is used to identify adulteration of yam decoction pieces to solve the problem of spectral fluctuations caused by the looseness of powder samples.The accuracy of the test set is 100% by applying the MSCIGA-SVM model in the identification of yam,winged yam and cassava pieces,which can also be used in adulteration of similar medicinal materials.Origin traceability can be realized by using the MSC-IGA-SVM model to identify eight kinds of yam origin.The accuracy rate of the MSC-IGA-SVM model is 97.30%,which is 0.87% higher than the accuracy rate of the SVM model(96.43%)built by directly using the original signal.Moreover,the MSC-IGASVM model reduces the calculation time for each classification from several hours to about several minutes.Experimental results show that the accuracy and calculation efficiency of the MSC-IGA-SVM model in identifying yam pieces have been improved.(3)The whole process algorithm of classification was applied to solve the spectral fluctuations resulting from the uneven surface of plant leaves using in this paper.The LIBS spectrum of leaves is used to non-destructively identify the citrus sugar grade.The MSC-IGASVM model is adjusted to the IGA-SVM model to identify three levels of citrus sugar,which solves the problem of high-precision and non-destructive identification of citrus sugar levels.The accuracy of the test set is increased from 89.60% to 92.03% with the adjusted model.Compared with the SVM model with the best recognition effect of 75.56% using the original signal,the accuracy is increased by 14.04% and 16.47% respectively.Through the one-way analysis of variance to explore the significant influence of leaf elements on citrus sugar recognition,the results show that the Mg element in leaves has a significant effect on citrus sugar recognition.These results show that adjusting the MSC-IGA-SVM model can realize the accurate and non-destructive identification of citrus sugar grades through leaves.In summary,through theoretical analysis of the whole process algorithm of LIBS classification,this paper selects the MSC-IGA-SVM model which can solve the problems of spectral volatility,feature redundancy and low recognition rate.The validity of the MSC-IGASVM model was verified through the identificaton of adulterate yam decoction pieces and the non-destructive identification of citrus sugar grades.The results show that the model can effectively improve the accuracy and calculation efficiency,and can be adjusted according to actual applications.The algorithm research in this paper provides a new method for the practical application of LIBS.