Study On Laser Induced Breakdown Spectroscopy Method For Detecting Key Elements In Infant Formula

Calcium and potassium are important nutrient elements in infant formula.The content of calcium and potassium is directly related to the quality of infant formula.They are representative elements in infant formula.The current elements detection technologies have many limitations that could not meet the current needs.Therefore,it is urgent to develop novel and efficient detection method to detect the nutrient elements in infant formula,which has attracted more and more attentions in the field of dairy food safety.Among all the elements detection technologies,Laser induced breakdown spectroscopy technology is one of best methods for element detection because of its advantage of simplicity and efficiency.The capability of LIBS technology with the chemometrics algorithms would definitely present great potential in real time quantitative analysis of nutrient elements in infant formula.This paper tries to introduce LIBS technology into the field of nondestructive detection of calcium and potassium in infant formula.In this paper,a series of work was carried out on the rapid and nondestructive detection of calcium and potassium in the infant formula.A normalized sample preparation and detection parameters,that conclude the delay time and position of convergent lens,etc.,were carried out to achieve better spectrum data.The quality of LIBS spectrum data will be influenced by the scattering effect and Stark effect.In this paper,a novel strategy,named as data-driven LIBS(DD-LIBS),is proposed to solve this problem.DD-LIBS combine HDWT,EMSC and Random frog algorithms to get better signals from complicate LIBS spectrum data.To overcome the disadvantage of random frog algorithm,this paper proposed a Modified Random Frog algorithm(MRF).DD-LIBS strategy use data as the driving force to improve the detection accuracy and ignore the physical processed.DD-LIBS dynamically adjust the parameters of the algorithms used in spectral pretreatment to waken the scattering effect and Stark effect and get a better result in quantitative analysis.To verify the validity of DD-LIBS,80 milk powder samples produced by different countries were prepared and detected,where 53 samples were randomly selected as correction set,and the remaining 27 samples were set as prediction set.The linear correction coefficient(R~2)of the DD-LIBS models for prediction of the calcium and potassium content are 0.9521,0.9556,respectively.The mean relative root mean square errors are 5.3%and 4.95%,respectively.This indicated the detection accuracy of DD-LIBS accords with national and industry standard.In this paper,LIBS was successfully adopted for detection of calcium and potassium element in infant formula.DD-LIBS strategy was developed to suppress the scattering effect and Stark effect.The results show that DD-LIBS is a powerful tool for rapid and nondestructive detection of nutrient elements in infant formula.