In order to investigate the feasibility of spectral transfer between near infrared spectrometric(NIRS) models of grating type and Fourier transform near infrared analyzer

the spectral transfer of original spectrum was performed by piecewise direct standardization(PDS) method

with domestic fish meal as the NIRS sample

DS2500F as the source instrument and MPA as the target one.The NIRS quantitative analysis prediction models for moisture

crude protein

crude fat

methionine and lysine were established

whose spectrum difference

applicability and accuracy were verified by correlation coefficient of cross validation(R2cv)

root mean square error of cross validation(RMSECV)

Mahalanobis distance(MD)

bias

root mean square error of prediction(RMSEP) and relative prediction deviation(RPD)

respectively.According to the verification and evaluation results

the various parameters of the prediction models established for moisture

crude protein

crude fat

methionine and lysine were not significantly different from the NIR spectrum of DS2500F instrument transferring to the target spectrum of MPA instrument.It showed that the NIR spectrum of domestic fish meal on the DS2500F instrument could be used to replace the original MPA spectrum by transmission

and model transfer was thus realized indirectly

which improved the application efficiency of NIR prediction model with good shareability and applicability.