Research On Quantitative Proportions In Complex Systems Of Vegetable Blended Oils Based On Multiple Regression Model And Cloud Computing

Edible vegetable oil has already become people daily necessary as one kind of important food which can provide heat energy and essential fatty acids for human,and promote fat-soluble vitamins absorption.So,edible oil safety concerns the people’s health and quality of life.In addition,there are a series of problems in blended oil market,such as shoddy,optional blending,named symbol confusion.Therefore,the research of blend oil detection method is very important and necessary.These are not only two difficulties but also two hot topics.Blended oil,as a kind of product,which can make up for single nutrient of single variety edible oil,is very popular with people in daily life.However,there have not been national standards for blended oils supervision for a long time,especially to blend oil constitutive proportion,thus lead to some enterprises do on the margins of legality and to mislead consumers.Blend oil product’s formula and proportion test,is actually a qualitative and quantitative analysis.On the premise of known plant blend oil formula,this paper focused on quantitative analysis of constitutive proportion in blended oils.The following four aspects of work were done with the help of hyphenated chromatographic instruments and chemometric approaches.1.Triglycerides are the main ingredients in vegetable oils were firstly analyzed comprehensively,by hyphenated GC-FID and HPLC-RID methods(chapter 2).The quantitative analysis of the various triglyceride compositions was carried out by internal standard method.Systematic methodology validation was studied from the linear,limits of detection(LODs)and limits of quantification(LOQs),repeatability and reproducibility.The systematic and comprehensive measurement uncertainty evaluations of triglyceride content testing in the whole experimental process were completed combined with the national standard’s provision.Using the uncertainty research to find out main influence component,thus find weaken or eliminate the system error influenced on test results.2.Stability of triglyceride in oil is the most major premise and the core standard for quantitative analysis of constitutive proportion in blended oils.The stability of soybean oil,rapeseed oil,corn oil,sunflower oil and peanut oil have been evaluated by oil shelf life simulation via taking out oil samples regularly in low temperature oven and tracing the changes of their acid value,peroxide value and triglyceride(chapter 3).Furthermore,the influence of different refining processes and conditions on the content of triglyceride in the oils has been researched.With the above studies,stable marker has been determined,it provided new thoughts of research on quantitative analysis of the pure oils in vegetable blend oil.3.Fingerprint method can fully reflect the characteristic chromatogram of the chemical components in vegetable oil,in order to build up qualitative/quantitative analysis model based on the comprehensive specific chromatogram extracted from the fingerprints of different pure oils,it was necessary to set up standard chromatogram database(chapter 4),and quality control procedure for monitoring the data validation in the whole detection process was also needed.Meanwhile,more precise corrected values have been obtained by correction on multiple experimental data,it was a firm base of the big data platform for composition research of vegetable blend oil.Descriptive analysis has been performed on the samples in the database via SPSS,clustering effect of soybean oil,rapeseed oil,corn oil,sunflower oil and peanut oil has been studied with multi statistical analysis.With the method of gravity distance between classes,some speculations have been made for the quantitative analysis on the composition of the five oils in vegetable blead oil.4.In chapter 5,in the mathematical model based on Multiple Linear Regression and Constrained Least Squares,triglyceride content in pure oil and vegetable blend oil were characteristic dependent variables,pure oil proportions were independent variables.The source factors of maximum error have been evaluated by simulate computation,method of exhaustion with Matlab/Linux/Cloud Computing was used in the quantitative detection of the proportions in binary/ternary vegetable blend oil,a diversity of evaluation and analysis have been performed on the final result.Furthermore,especially for the vegetable blend oil system with coexistence of soybean oil and corn oil,the characteristic dependent variable with the combination of gas chromatography triglyceride and liquid chromatography triglyceride has been introduced for the first time,the prediction result with it was significantly better than the prediction result with gas chromatography triglyceride as the characteristic dependent variable.In the paper,with close integration of chromatographic analysis on triglyceride in vegetable blend oil,chemometrics and big data calculation,some fundamental research on soybean oil,rapeseed oil,corn oil,sunflower oil,peanut oil and binary/ternary vegetable blend oil with the 5 oils have been done,the advantage and the huge potential of this integration on quantitative detection of the proportions in vegetable blend oil,quality control and other areas were fully demonstrated.