The Research And Application Of SIMCA Method Based On Infrared Spectroscopy

With the development of computer technology,the acquisition of spectral is becoming easier and easier,and the spectral has been called the ideal information carrier of analytical technology.Spectral analysis technology combines the spectral and pattern recognition methods to achieve the qualitative and quantitative analysis of complex systems.It is suitable for laboratory routine analysis,field analysis and on-line analysis.It has been widely used in petrochemical,medicine,agriculture and food detection with great development potential and bright application prospects.Although the application of spectral pattern recognition method is mature but it also has flaws.In this paper,the spectral qualitative analysis method is studied and applied.SIMCA method is the most widely used and the most mature method in qualitative analysis.In the SIMCA,the parameters of PC A model and F test are used to construct T2 and Q for classification,and Euclidean distance is used to determine the range of samples distribution of the model.Because the range defined by Euclidean distance,which is a circle in the plane of T2 vs Q,can not represent accurately the boundary of actual samples which distributes in some directions and irregular space,SIMCA is still inaccurate for classification and identification in theory.A new multivariate classification and identification method using Mahalanobis Distance instead of Euclidean distance was proposed in this thesis.Experiments of infrared spectra of blending edible oils and near infrared spectra of animal furs were designed to compare the performance of the new method and SIMCA.The distribution range of the samples was calculated by Mahalanobis distance and Euclidean distance show that the classification and recognition ability of Mahalanobis distance were stronger.And the results of the validation set show that the new method has better accuracy than the SIMCA in the classification of chemical composition with little difference.At present,the adverse effects of edible oil,such as adulteration,deterioration,and bad taste of edible oil,have had a bad influence on the society.Infrared spectroscopy has the advantages of simple operation,rapid analysis,no pollution,no damage,etc.It is suitable for fast detection and on-line analysis of edible oil.In this paper,the qualitative analysis for edible oil by infrared spectroscopy and chemometrics was proposed.The composition of edible oil is similar and the SIMCA classification method we used in this area are not ideal,a kind of improved method was put forward.Based on the PCA model,SVDD was used to determine the type of data.SVDD has 2 adjustable parameters so it can be adjusted with the distribution of the data to make the results more convincing.The edible oil and blend oil experimental results show that SIMCA method can not distinguish oil and other oil,while the improved SIMCA method can.In the classification of different blend oil,two kinds of methods can not completely distinguish the categories,but the improved method has higher classification accuracy,which could provide a new method to solve the problem of edible oil on-line analysis.