基于近红外光谱与BP神经网络预测落叶松木屑的含水率

利用近红外光谱(NIR)技术结合BP神经网络定量预测了落叶松木屑的含水率。首先对采集的落叶松木屑原始近红外光谱进行9点平滑及多元散射校正预处理,然后利用主成分分析法提取光谱数据主成分作为BP神经网络的输入,最后建立BP神经网络预测模型并采用交叉验证法对模型进行验证。所建模型校正集的相关系数R为0.98,校正集的均方根误差RMSEC为0.001 7;预测集的相关系数R为0.99,预测集的均方根误差RMSEP为0.001 5。研究表明,此方法可以实现对落叶松木屑含水率的快速预测。

Prediction of Sawdust Water Content of Dahurian Larch Based on NIRS and BP Neural Network

An integration of BP neural network and PCA for modeling sawdust water content of Dahurian Larch combined with NIRS was investigated in this paper. The raw spectra were collected and pretreated with 9 point smoothing and multiplieative scatter correction （MSC）. Two typical principal components were extracted by PCA with the application of establishing prediction model. U- sing full cross-validation approach to validate the model, the calibration correlation coefficient （R） was 0. 98 and the root mean square error of calibration （RMSEC） was 0. 001 7, the prediction correlation coefficient （R） was 0. 99 while the root mean square error of prediction （RMSEP） was 0. 001 5. The study results showed that this method can rapidly and accurately predict sawdust water content of Dahurian larch.