A Study On Classification And Recognition Of Sesame Oil Based On Terahertz Spectroscopy

Sesame oil is a nutritious and very delicious food seasoning material.The quality of sesame oil is a very important problem.Since the conventional detection method requires a certain detection time,the process is complicated.Therefore,in view of these shortcomings,this paper explores a rapid and convenient detection method,which is based on the terahertz time-domain spectroscopy system and combined with statistical analysis methods to detect sesame oil.Terahertz detection has the advantages of fast,non-destructive,simple operation,etc.Therefore,the research on the quality of sesame oil can quickly obtain the test results.The main research contents of this paper are as follows:Firstly,this paper uses terahertz time-domain spectroscopy system to detect ten different sesame oils,analyzes the spectrum,calculates absorption coefficient and refractive index,preprocesses the data by principal component analysis and then builds the model through support vector machine.Differentiate and identify different sesame oils.According to the characteristics of sesame oil in the terahertz spectrum,the support vector machine model is optimized by different parameter optimization algorithms.The mean square error is used as a criterion for measuring the predictive performance of the model.According to the prediction results,the support vector machine model constructed in this paper has a good classification and recognition effect,and compared with other methods,the conclusion that the model has more accurate recognition results is obtained.Secondly,this paper also analyzes the change of time domain spectrum by adding 70% of other kinds of sesame oil to the same sesame oil,and analyzes the absorption spectrum calculated by the change in the time domain to analyze the adulterated sesame oil and pure sesame oil in terahertz.Features presented in the spectroscopy system.According to these characteristics,the quality analysis of adulterated sesame oil and pure sesame oil was further studied to determine whether sesame oil was incorporated into other oils,to carry out two-class research,and to evaluate the model by TP,TN,FP,FN and other evaluation indicators.The classification ability is judged,and the best model is combined with the main component analysis and the support vector machine.Finally,the paper prepared a mixture of sesame oil with 10%,20%,30%,40%,50% soybean oil,and then detected it through terahertz time-domain spectroscopy to study the blending in sesame oil.When different proportions of soybean oil were added,the quality changes showed different in the terahertz time-domain spectrum,and the difference was analyzed by the partial least squares algorithm to find the difference between the ratio and the absorption coefficient of the incorporated soybean.Based on this,a partial least squares quantitative research model of sesame oil adulteration was constructed based on this.Quantitative analysis of adulterated sesame oil by partial least squares model showed a good fitting effect.