As some existed influencing factors resulting from the precise and the measure environments of spectrum instrument, the spectrum drifts, the linear or nonlinear conversion can be found in the process of the measure. Prediction errors will be made if we directly apply the results of the sample's measurement in the construction and prediction of the model. Therefore, multivariate calibration algorith——Shenk's algorithm was adopted to correct the discrepancy of the erucic acid near-infrared spectrums and it was indicated that predication precise would be improved greatly, root mean square errors of prediction (RMSEP) has been reduced to 1.569 from 2.263, and average relative error was dropped to 3.218% from 4.6%, with the relative coefficient improved to 0.913 from 0.780.
參考文獻
相似文獻
引證文獻
引用本文
陳斌,吳繼明.用Shenk's方法提高芥酸NIR模型預測精度[J].農(nóng)業(yè)機械學報,2007,38(8):101-104.[J]. Transactions of the Chinese Society for Agricultural Machinery,2007,38(8):101-104.