Detection Of Mutton Adulteration Based On Fiber Optic Spectrum Technology

Lamb meat is delicious and rich in nutritional value.It is deeply loved by consumers and its demand at home and abroad continues to grow.However,because many mutton processing companies are difficult to manage,it provides opportunities for some illegal vendors,which has led to the intensification of mutton adulteration.In order to protect the rights and health of consumers,meat adulteration testing is necessary.At present,some chemical methods have been used to detect whether meat is adulterated,but these methods have certain defects,which damage the meat and cause a certain waste of time or economy.In order to make up for the lack of chemical methods,the paper uses fiber optic spectroscopy to adulterate lamb.In-depth research on detection methods.The main work of this paper is as follows:(1)Pretreatment of spectral data.Firstly,the spectral reflectance of the adulterated mutton samples was obtained based on the optical fiber spectroscopy technology.Then,the original spectral data was preprocessed,and the Savitzky-Golay convolution smoothing method was adopted to reduce the noise of spectral data,smooth the spectral curve,and improve the signal to noise ratio of data.The data with different dimensions are normalized to the interval[0,1]to make the data comparable.The data were provided for the establishment of mutton adulteration detection model.(2)A qualitative detection model of mutton adulteration based on weighted random forest algorithm was established.First sample can be divided into two categories:adulteration pure samples of mutton and mutton samples,adulterated mutton samples including "lamb-chicken" and "lamb-pork" adulterated samples,established the mutton adulterated detection model based on random forest algorithm,identify adulteration sample is pure lamb or mutton,results show that the model accuracy is 96.67%.Then further to identify mutton adding meat,sample can be divided into three categories:pure samples of mutton,"lamb-chicken" adulterated sample and "lamb-pork" adulterated samples,established a mixed type detection model based on random forest algorithm,identify the types of lamb meat adulterated in,the results show that the model of the classification accuracy of 93.33%.In order to improve the accuracy of the model,the weighted method was used to optimize the voting process of the random forest algorithm,and the incorporation type detection model based on the weighted random forest algorithm was established.The results showed that the classification accuracy of the model was 96.67%,which was 3.34%higher than that of the traditional random forest model.(3)A quantitative detection model for mutton adulteration based on particle swarm optimization least-squares support vector machine was established.The adulterated samples of "mutton and pork" were selected to detect the content of pork adulterated in mutton.Feature wavelength was extracted from spectral data by competitive adaptive reweighting method and monte carle-uninformative variable elimination method.Full-wavelength model and feature wavelength model were established based on the least squares support vector machine algorithm.The results showed that CARS-LSSVM model had the highest accuracy in detecting the pork concentration in mutton,with R2=0.9750 and RMSE=0.0508.Then particle swarm optimization was used to optimize the two parameters of the CARS-LSSVM model,and the results showed that R2=0.9930 and RMSE=0.0269.Compared with the unoptimized CARS-LSSVM model,R2 and RMSE were improved by 0.018 and 0.0239 respectively.