Study On Quantitative Detection Method Of Adulterated Meat Based On Electronic Nose

There are some drawbacks in the meat market,such as meat adulteration.Lawless producers blend expensive meat with cheap meat,such as by blending beef with pork to increase profits.Besides the economic loss,the illegal activity of meat adulteration also harms people with certain religious beliefs.Related technologies for the detection of meat adulteration have been developed,such as biology-based technology and chemistry-based technology,however,these technologies are expensive and time-consuming.An electronic nose(E-nose)is a chemical measurement system used to measure the chemical properties of volatile gases and has been widely applied to detect the quality and safety of food due to its fast speed,high reliability,simple operation,and relatively low cost.In this study,a PEN3 E-nose is used for the quantitative detection of minced beef adulterated with pork.Previous related studies manually selecting and extracting features of E-nose data.These manual operations not only burden the user with complex tasks but are also likely to lose important information.To improve the precision of detected results,a 1DCNN-RFR framework is proposed to automatically extract features of E-nose data.Firstly,the meat data with 7 different adulterated proportions(0%,10%,20%,30%,40%,50%,and 60%)are collected based on the PEN3 E-nose.Then,traditional machine learning methods(MLR、SVR、RF、BPNN),combined with stable values,are used to predict the adulterated proportions.However,the performances of these traditional machine learning methods are unsatisfactory.Finally,the proposed 1DCNN-RFR framework is used to predict the adulterated proportion.The results indicate that the proposed1DCNN-RFR framework has good performance.This study indicates that the proposed 1DCNN-RFR framework could be used as an effective tool for the quantitative detection of meat adulteration.