近红外光谱分析技术在土壤含水率预测方面具有独特的优势, 是一种便捷且有效的方法。 卷积神经网络作为高性能的深度学习模型, 能够从复杂光谱数据中自主提取有效特征结构进行学习, 与传统的浅层学习模型相比具有更强的模型表达能力。 将卷积神经网络用于近红外光谱预测土壤含水率, 并提出了有效的卷积神经网络光谱回归建模方法, 简化了光谱数据的预处理要求, 且具有更高的光谱预测精度。 首先对不同含水率下土壤样品的光谱反射率数据进行简单的预处理, 通过主成分分析减少光谱数据量, 并将处理后的光谱数据变换为二维光谱信息矩阵, 以适应卷积神经网络特殊的学习结构。 然后基于卷积神经网络算法, 设置双层卷积和池化结构逐层提取光谱数据的内部特征信息, 并采用局部连接和权值共享减少网络参数、 提高泛化性能。 通过试验优化网络结构和各项参数, 最终获得针对土壤光谱数据的卷积神经网络土壤含水率预测模型, 并与传统的BP, PLSR和LSSVM模型进行对比实验。 结果表明在训练样本达到一定数量时, 卷积神经网络的预测精度和回归拟合度均高于三种传统模型。 在少量训练样本参与建模的情况下, 模型预测表现高于BP神经网络, 但略低于PLSR和LSSVM模型。 随着参与训练样本量的增加, 卷积神经网络的预测精度和回归拟合度也随之稳定提升, 达到并显著优于传统模型水平。 因此, 卷积神经网络能够利用近红外光谱数据对土壤含水率做出有效预测, 且在较多样本参与建模时取得更好效果。

The technology of near infrared spectroscopy that has unique advantage in the prediction of soil moisture content is a convenient and effective method. Convolutional neural network (CNN) is a deep learning model with high performance. Using CNN, effective features data can be extracted from complex spectral data and the inner structure of feature data can be learned. Compared with traditional surface learning models, convolutional neural network has more powerful modeling capability. In this research, the CNN was used to predict the soil moisture content by near infrared spectroscopy. An efficient modeling method of CNN for spectral regression was proposed. The pretreatment process of spectral data was simplified and the accuracy of spectral prediction was improved by this modeling method. In this paper, firstly, the simple pretreatment was used to treat the spectral reflectance data of soil samples under different moisture contents. Principal component analysis was used to reduce the amount of spectral data and the correlation of the features. The processed spectral data was transformed into 2-dimensional spectral information matrixes to meet the special learning structure of CNN. Secondly, the convolutional neural network was used to build the regression model for the prediction of soil moisture content. The first four stages of this CNN model were composed of two types of layers: convolutional layers and pooling layers. Inner features of the input spectral data were obtained by composing convolutional layers and pooling layers, each transforms the representation at one level into a representation at a higher, slightly more abstract level. With the composition of enough such transformations, very effective inner features of spectral data can be extracted. There were two key ideas behind the CNN model that can reduce the number of parameters of the network: local connections and shared weights. In addition, these ideas can also improve the generalization performance of the CNN model. The model structure and parameters were optimized by carrying out experiments. Finally, the CNN model with improved regression structure of soil spectral data was built for the prediction of soil moisture content. The CNN model was compared with the BP, PLSR and LSSVM models, and these three traditional models were commonly used in the prediction of soil moisture content. The results showed that when the number of training samples reached to some degree, the prediction accuracy and regression fitting degree of the CNN model were higher than those of the traditional models. The performance of the CNN model were much higher than the BP neural network which had the same network training method with the CNN model, but slightly lower than the PLSR and LSSVM models when a small number of training samples were used in the modeling. The prediction accuracy of the CNN model greatly increased with the number of training samples growing. So did the regression fitting degree of the CNN model. In the end, the performance of the CNN model was significantly better than the traditional models. Therefore, the CNN method could be used to effectively predict the soil moisture content by the near infrared spectral data, and better results are obtained when more training samples are involved in modeling.