Rapid Screening Of Contaminated Peanut Oil And Design Of Online Platform Of Analysis Of Infrared Spectra

Peanut oil is one of the most important edible oil in daily life.It is known as"the first oil of the East"and is closely related to the health of the people and the international economic and trade.Aflatoxins is the notorious hazard for peanut oil industry.Therefore,rapidly screening of peanut oil with aflatoxins exceeding limit is essential.In this paper,chemometrics combined with FTIR spectroscopy were used for rapidly screening of peanut oil with aflatoxins?AFT?or aflatoxin B1?AFB1?over standards.The standards are in conformity with standards of China?AFB1<20ppb?and the US?AFT<20ppb?.Firstly,the spectral variation of contaminated peanut oil samples were interpreted in the mid infrared spectral region,the aflatoxins concentration is far lower than the glycerol content of fatty acid,the absorption peaks were very weak and just completely overlapped by absorption peaks of glycerol,which was very difficult to observe apparently.Hopefully,the obvious phenomenon of hydrolysis was a result of aflatoxins intensification in oil,variation of absorption peaks around 1760-1680 cm^-1 appeared.The?Absorbance ratio-Savitzky Golay preprocessing-partial least square?AR-SGPLS,?Absorbance ratio-principal component analysis-linear discriminant analysis?AR-PCALDA and?Absorbance ratio-moving window-linear discriminant analysis?AR-MWLDA methods were applied to this region to develop classification models,all results were compared to find optimal model with less spectral data points involved.AR-MWLDA combined with the window adjustment was developed as optimal model with only 22 spectral points involved modeling.Calibration and validation sets in model could reach 100%true positive rate and true negative rate.This result could make it more efficient and economical to design multi-task instrument for screening aflatoxins and aflatoxin B1 contaminated peanut oil.Furthermore,online infrared spectra analysis platform with the model-sharing and self-updating was designed based on JAVA WEB,HTML,MYSQL,etc.The design of user interface,algorithm integration,business logic,establishment of database were elaborated.Brief description of relevant theories and technologies,system architecture and functions of modules were carried out.The functions of the platform were preliminary tested,expectation was achieved successfully.The achievement of platform provided an access to sharing of models and a good approach of gathering data.Further improvement of the system were needed,great contributions would also make progress of the industry of infrared spectral analysis.