基于注意力机制的BiGRU土壤光谱全氮预测模型研究

doi:10.3964/j.issn.1000-0593(2025)07-2017-09

摘要
参考文献
相关文章 (15)

全文: PDF (8185 KB)
输出: BibTeX | EndNote (RIS)

摘要：土壤全氮含量是评估土壤肥力的关键指标，其精确测定对于提升农作物产量和品质具有重要意义。运用近红外光谱分析技术预测土壤全氮含量已被证明是一种有效的解决方案。由于土壤光谱数据具有高维性和复杂的时间序列性，传统模型往往难以捕捉其中的关键信息，从而影响预测结果的准确性。为此基于600份土壤样本的近红外光谱（900～1 700 nm），开展了土壤全氮（STN）含量光谱预测方法研究，提出了一种基于注意力机制的双向门控循环单元模型（BiGRU-Attention）。首先通过SG滤波和SNV预处理方法优化了光谱数据，随后通过CARS特征筛选算法将光谱的波长数由198精简为30个关键特征波长，剔除冗余信息，降低了建模的复杂度。BiGRU-Attention模型利用更新门和重置门有效控制信息流动，使得模型忽略不重要的光谱数据，并保留影响预测精度的关键信息。通过结合双向GRU的双时序处理优势，模型能够同时处理光谱序列的正向与反向输入，从而增强模型对边缘数据的关注能力，更全面地捕捉土壤光谱数据中的前后依赖关系。此外，模型通过注意力层的QKV矩阵计算每个部分的重要性，并根据序列中的前后关联信息动态决定关注哪些特征，通过计算注意力权重矩阵，为每个输入数据分配权重，生成更相关的上下文矩阵，进而增强模型的预测精度。实验结果表明，与其他模型相比，BiGRU-Attention模型能更好地理解波段之间的相互关联，在预测结果上表现更佳，光谱数据在经过特征筛选后，模型在测试数据集上的决定系数R²达到了0.87，均方根误差RMSE为0.20 g·kg^-1，表现出良好的预测性能。该研究为土壤养分快速检测提供了技术支持，为建立高精度的土壤全氮含量预测模型提供了方法与参考。

关键词：近红外光谱；双向门控循环单元；注意力机制；土壤全氮

Abstract：Soil total nitrogen (STN) content is a key indicator for evaluating soil fertility, and its accurate measurement is of great significance for improving crop yield and quality. Predicting soil total nitrogen content using near-infrared spectroscopy has been proven to be an effective solution. However, due to the high dimensionality and complex time series characteristics of soil spectral data, traditional models often struggle to capture critical information, affecting prediction accuracy. Hence, based on near-infrared spectra (900～1 700 nm) of 600 soil samples, the method for predicting soil total nitrogen (STN) content was investigated,and a Bidirectional Gated Recurrent Unit based on an Attention Mechanism (BiGRU-Attention) model was proposed. First, the spectral data quality was optimized using SG filtering and SNV preprocessing methods. Then, using the CARS feature selection algorithm, the wavelengths for modeling were reduced from 198 to 30, thereby removing redundant information and decreasing the complexity of the modeling process. The BiGRU-Attention model effectively manages the flow of information using update and reset gates and enables the model to disregard the unimportant spectral data and retain the key information, which impacts the prediction accuracy. By leveraging the dual temporal sequence processing advantages of bidirectional GRU, the model can simultaneously handle forward and backward inputs of spectral sequences, thereby enhancing its ability to focus on edge data and comprehensively capture the dependencies present in soil spectral data. Additionally, the model employs an attention layer to compute the importance of each segment using the QKV matrix and dynamically determines which features should be emphasized based on the sequential interdependencies. This process calculates attention weight matrices to assign weights to each input data point, generating a more relevant context matrix that improves the model's predictive accuracy. Experimental results show that the BiGRU-Attention model can better understand the correlation between bands and perform better in prediction than other models, with the spectral data achieving an R² of 0.87 and an RMSE of 0.20 g·kg^-1 on the test dataset after feature selection. This study provides technical support for rapid soil nutrient detection and offers a method and reference for establishing high-accuracy STN prediction models.

Key words：Near-infrared spectroscopy; Bidirectional gated recurrent unit; Attention mechanism; Soil total nitrogen

收稿日期: 2024-09-23 修订日期: 2025-02-25

中图分类号:

TP183

基金资助: 国家重点研发计划项目（2023YFD1701000)资助

通讯作者: 李民赞 E-mail: limz@cau.edu.cn

作者简介: 剧伟良，1996年生，中国农业大学智慧农业系统集成研究教育部重点实验室硕士研究生 e-mail: s20223081703@cau.edu.cn

引用本文:

剧伟良，杨玮，宋亚美，刘楠，李浩，李民赞. 基于注意力机制的BiGRU土壤光谱全氮预测模型研究[J]. 光谱学与光谱分析, 2025, 45(07): 2017-2025.
JU Wei-liang, YANG Wei, SONG Ya-mei, LIU Nan, LI Hao, LI Min-zan. Prediction of Soil Total Nitrogen Based on NIR Spectroscopy and BiGRU Model With Attention Mechanism. SPECTROSCOPY AND SPECTRAL ANALYSIS, 2025, 45(07): 2017-2025.

链接本文:

https://www.gpxygpfx.com/CN/10.3964/j.issn.1000-0593(2025)07-2017-09 或 https://www.gpxygpfx.com/CN/Y2025/V45/I07/2017

[1] BAI Zi-jin, PENG Jie, LUO De-fang, et al（白子金，彭杰，罗德芳，等）. Spectroscopy and Spectral Analysis(光谱学与光谱分析), 2022, 42(9): 2768.
[2] GAO Fei, CHEN Su-xian, DU Xiao-yu, et al（高飞，陈素贤，杜晓玉，等）. Tianjin Agricultural Sciences(天津农业科学), 2022, 28(9): 76.
[3] CHEN Hao-yu, XIANG Lei, GAO He, et al（陈昊宇，项磊，高贺，等）. Remote Sensing for Natural Resources (自然资源遥感), 2023, 35(3): 170.
[4] ZHANG Xi-yu, LI Si-jia, WANG Xiang, et al（张锡煜，李思佳，王翔，等）. Transactions of the Chinese Society of Agricultural Engineering(农业工程学报), 2023, 39(15): 144.
[5] ZHANG Ke-di, GUO Hong-yun, SHAO Na, et al（张珂迪，郭洪云，邵娜，等）. Journal of Anhui Agricultural University(安徽农业大学学报), 2023, 50(6): 917.
[6] HUANG Ye-chang, ZENG Wei, DENG Li-zhang, et al. （黄业昌，曾玮，邓力章，等）. Journal of Zhejiang Agricultural Sciences (浙江农业科学), 2024, 65(3): 536.
[7] ZHOU Xu-dong, SHEN Yun-xin, PU Yong-yu, et al（周旭东，申云鑫，濮永瑜，等）. Jiangsu Agricultural Sciences (江苏农业科学), 2024, 52(4): 231.
[8] TANG Ao-hua, YANG Gui-jun, YANG Yue, et al（唐澳华，杨贵军，杨悦，等）. Transactions of the Chinese Society for Agricultural Machinery(农业机械学报), 2024, 55(8): 382.
[9] Metzger K, Bragazza L. European Journal of Soil Science, 2024, 75(3): e13508.
[10] Sitorus A, Lapcharoensuk R. Sensors, 2024, 24(7): 2362.
[11] ZHOU Peng, KONG Yi-nuo, HAO Shan-shan, et al（周鹏，孔一诺，郝珊珊，等）. Transactions of the Chinese Society of Agricultural Engineering(农业工程学报), 2024, 40(16): 113.
[12] WEI Qing, LI Chang-yu, XU Meng-cao, et al（卫青，李长昱，许孟操，等）. Journal of Anhui Agricultural Sciences (安徽农业科学), 2023, 51(8): 6.
[13] Khammao P, Rattanapichai W, Pitakdantham R, et al. Soil Science and Plant Nutrition, 2024, 70(3): 197.
[14] QIN Wen-hu, DONG Kai-yue, DENG Zhi-chao（秦文虎，董凯月，邓志超）. Soils(土壤), 2023, 55(6): 1347.
[15] HUANG Hai-xia, LI Yao-xiang, ZHANG Zhe-yu（黄海霞，李耀翔，张哲宇）. Forest Engineering(森林工程), 2023, 39(6): 164.
[16] Tan A, Wang Y, Zhao Y, et al. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 2022, 283: 121759.
[17] Bai L, Zhang Z T, Guan H, et al. Talanta, 2024, 275: 126098.
[18] Zeng W, Qiu Y, Huang Y, et al. Journal of Innovative Optical Health Sciences, 2022, 15(4): 2250022.
[19] Bai Z, Chen S, Hong Y, et al. Geoderma, 2023, 437: 116589.
[20] Wang Y, Chen S, Hong Y, et al. Computers and Electronics in Agriculture, 2023, 212: 108067.
[21] Yuan Q, Wang J, Zheng M, et al. Construction and Building Materials, 2022, 350: 128799.
[22] Ji X, Huang H, Chen D, et al. Buildings, 2023, 13(1): 72.
[23] MAO Bai-hai, QIN Wu, XIAO Xian-bin, et al(毛百海，覃吴，肖显斌，等). Energy Storage Science and Technology(储能科学与技术), 2023, 12(11): 3519.
[24] Pan S, Yang B, Wang S, et al. Energy, 2023, 284: 128701.
[25] Guo X, He J, Wang B, et al. Remote Sensing, 2022, 14(19): 4737.
[26] Kong F, Li J, Jiang B, et al. IEEE Transactions on Industrial Informatics, 2023, 19(1): 541.
[27] WANG Wei-chao, YANG Wei, CUI Yu-lu, et al（王炜超，杨玮，崔玉露，等）. Transactions of the Chinese Society for Agricultural Machinery(农业机械学报), 2021, 52(S1): 316.
[28] CAO Yong-yan, YANG Wei, WANG Dong, et al（曹永研，杨玮，王懂，等）. Transactions of the Chinese Society for Agricultural Machinery (农业机械学报), 2022, 53(S1): 241.