Study On Laser Induced Breakdown Spectroscopy Database And Classification Method Of Plastic Samples

Laser induced breakdown spectroscopy(LIBS)has the advantages of multi-element simultaneous analysis?no sample preparation,rapid speed,small sample damage and no influence by the sample color.Combined with pattern recognition method based on chemometrics can be used for sample classification,which can extract feature information of plastic samples,achieve better classification accuracy and shorter classification time.However,the current classification models based on the existing classification method has the disadvantages of too many parameters to be adjusted and universality of the models.So,the models are difficult to apply in practical applications.The classification model based on Partial least squares discriminate analysis(PLS-DA)can overcome the background interferences and improve classification accuracy.The classification method combined with plastic retained database can be used to identify plastic samples.However,the current LIBS database based on plastic samples has not been reported.And the basic database can not meet application requests.So,the establishment of exclusive database can help researchers apply LIBS in plastic areas.The main research contents are as follows:(1)The research of classification method based on LIBS and development of software.The LIBS spectra is a complex atomic emission spectrum.Different samples have different element types and content.So,the characteristic peaks of their spectra have different wavelength locations and intensity.Those differences can be extracted by the pattern recognition method based on chemometrics which can realize the recognition of sample types.The sample classification model based on PLS-DA can improve the predictive ability of the model and develop sample classification software based on Matlab.The study of sample classification method can provide theoretical basis for subsequent practical applications.(2)Development of LIBS database for plasticThe database consists of a LIBS element characteristic spectral database and a proprietary database of plastic samples.LIBS uses the permanent line in NIST ASD database as the data source to establish the LIBS characteristic spectral line database,which contains 2058 characteristic lines of 99 elements;Based on the LIBS spectral data of 11 kinds of plastic samples as data source,A proprietary database of plastic samples was set up.The database has the ability to add?delete?query and modify.The establishment of LIBS database is the basic condition for qualitative and quantitative analysis.(3)Study on PLS-DA classification method for plastic samples based on LIBS.Using the LIBS device to experiment with 11 kinds of plastic samples,2200 spectra were obtained,and each spectrum was an average of 20 times.According to the PLS-DA algorithm,the prediction model is established,and the plastic is classified and identified.The results are analyzed,including the optimal proportion of training set and validation set,the optimal principal component number and error analysis of the model.The experimental results show that the classification accuracy of the verification set reached 99.80% and the classification accuracy of the test set reached 99.09% without replacing the interference spectrum.After replacing the interference spectrum,the sample classification accuracy of the verification set and the test set reached 100%,and the classification time has decreased.It can be concluded that the combination of LIBS and PLS-DA algorithm can realize the classification and recognition of plastic samples.