Fast Nondestructive Detection Of Edible Oil By Laser Spectrum Based On Band Screening

Edible oil is one of the essential cooking materials of Chinese cuisine,which is renowned all over the world.However,at present the market is flooded with a large number of adulterated and inferior cooking oil,mixing inferior oil into high-quality oil,seriously endangering the legitimate interests of the public.Due to the limitations of experimental conditions and materials,traditional edible oil detection methods need a lot of experiments and complex analysis,and are time-consuming.In view of this,this paper proposes a fast recognition model of edible oil fluorescence spectrum based on band screening and classifier.The rapid response of laser-induced fluorescence technology and the classifier algorithm based on band screening were used to identify and classify edible oil,which can achieve rapid detection and recognition.This topic on the analysis of the market of edible oil adulterated means,respectively select LuHua brand peanut oil,fook lam moon,fook lam moon corn oil and soybean oil LuHua brand rapeseed oil as experimental samples of material,the experimental process separately collected the fluorescence spectrum of the single oil,peanut oil,such as proportion of mixed proportion of fluorescence spectrum,such as peanut oil soybean oil mixed with corn oil percentage of fluorescence spectrum,such as peanut oil canola oil fluorescence spectrum,a total of 7 groups of experimental samples,in order to verify the exactness of the model for different kinds of edible oil spectrum identification.100 groups of fluorescence spectra were collected for each sample,and a total of 700 groups of fluorescence spectra were collected.Then,dimension reduction algorithm was used to select the fluorescence spectrum features and to reduce the dimension of the data.This can limit the redundant information in the spectrum and improve the accuracy and speed of recognition.For dimension reduction after data respectively in combination with support vector machine(SVM),extreme learning machine(ELM),random forests(RF)model training,through the horizontal comparison of the several kinds of classification algorithm research,found that the dimension reduction after the data combination of different classification algorithms have their respective advantages,and accuracy reached 100%,at the same time has a good generalization performance.In this paper,dimension-reduction operation combined with classification and recognition algorithm is introduced into the detection of edible oil for the first time,and laser induced fluorescence technology is used.Through the simulation of the model,it can be found that the establishment of the oil sample model is worthy of further study.This paper builds a fast and accurate identification model of edible oil,which can realize the rapid detection of edible oil.It also provides some reference for the establishment of a more perfect database model of laser-induced fluorescence spectrum of edible oil.Figure[40]table[5]reference[69]...