[1]李星宇,李玥,高玉红.基于CNN-BiLSTM-Attention模型的胡麻产量预测[J].江苏农业学报,2025,(07):1342-1349.[doi:doi:10.3969/j.issn.1000-4440.2025.07.010]
　LI Xingyu,LI Yue,GAO Yuhong.Flax yield prediction based on CNN-BiLSTM-Attention model[J].,2025,(07):1342-1349.[doi:doi:10.3969/j.issn.1000-4440.2025.07.010]
点击复制

基于CNN-BiLSTM-Attention模型的胡麻产量预测()
分享到：

江苏农业学报[ISSN:1006-6977/CN:61-1281/TN]

卷:
期数:
2025年07期
页码:
1342-1349
栏目:
农业信息工程
出版日期:
2025-07-31

文章信息/Info

Title:
Flax yield prediction based on CNN-BiLSTM-Attention model
作者:
李星宇1; 李玥1; 2; 高玉红2; 3
(1.甘肃农业大学信息科学技术学院,甘肃兰州730070；2.甘肃农业大学省部共建干旱生境作物学重点实验室,甘肃兰州730070；3.甘肃农业大学农学院,甘肃兰州730070)
Author(s):
LI Xingyu1; LI Yue1; 2; GAO Yuhong2; 3
(1.College of Information Science and Technology, Gansu Agricultural University, Lanzhou 730070, China；2.State Key Laboratory of Crop Science in Arid Habitats, Gansu Agricultural University, Lanzhou 730070, China；3.College of Agriculture,Gansu Agricultural University, Lanzhou 730070, China)
关键词:
胡麻; 产量预测; 深度学习; 卷积神经网络; 双向长短期记忆模型
Keywords:
flax; yield prediction; deep learning; Convolutional Neural Network; Bidirectional Long Short-Term Memory model
分类号:
S565.9
DOI:
doi:10.3969/j.issn.1000-4440.2025.07.010
文献标志码:
A
摘要:
本研究提出了一种用于胡麻产量预测的基于深度学习方法的卷积神经网络(CNN)-双向长短期记忆网络(BiLSTM)-注意力机制(Attention)模型,该模型整合了卷积神经网络的空间特征提取能力、双向长短期记忆网络的时序动态建模能力以及注意力机制的特征自适应加权功能。基于气候数据、植被指数和2000-2020年产量对模型进行训练。试验结果表明,CNN-BiLSTM-Attention模型预测精度显著优于传统模型,其均方根误差(RMSE)达到316.98 kg/hm2,决定系数(R2)达到0.83。该模型在年际气候变化条件下保持了良好的稳定性和较高的精确度。本研究为胡麻产量预测提供了技术支持,其模块化设计框架还可推广应用于其他作物的生长监测与产量预估。
Abstract:
This study proposed a deep learning-based model integrating Convolutional Neural Network (CNN), Bidirectional Long Short-Term Memory (BiLSTM) and attention mechanism to predict flax yield. The model combined the spatial feature extraction capability of CNN, the temporal dynamic modeling ability of BiLSTM, and the feature adaptive weighting function of the Attention mechanism. The model was trained using climate data, vegetation indices, and yield data during 2000-2020. Experimental results showed that the CNN-BiLSTM-Attention model significantly outperformed traditional models in prediction accuracy, with a root mean square error (RMSE) of 316.98 kg/hm2 and a coefficient of determination (R2) of 0.83. The model maintained good stability and high accuracy under interannual climate change conditions. This study provides technical support for flax yield prediction, and its modular design framework can also be extended to the growth monitoring and yield prediction of other crops.

参考文献/References:

[1]赵玮,党占海,李闻娟. 强抗旱胡麻新品种陇亚11号组织培养技术优化研究[J]. 中国沙漠,2012,32(5):1355-1361.
[2]XIE D W, DAI Z G, YANG Z M, et al. Genomic variations and association study of agronomic traits in flax[J]. BMC Genomics,2018,19:1-12.
[3]BOGACHEV A A, GAVRILOVA N A, KURDYUKOV E E, et al. Comparative study of component and fatty-acid composition of flax seeds[J]. Food Chemistry,2020,310:125934.
[4]ZIETKIEWICZ E, RAFALSKI A, LABUDA D. Genome fingerprinting by simple sequence repeat (SSR)-anchored polymerase chain reaction amplification[J]. Genomics,1994,20(2):176-183.
[5]张雯丽. 中国特色油料产业高质量发展思路与对策[J]. 中国油料作物学报,2020,42(2):167-174.
[6]崔红艳,方子森,牛俊义. 胡麻栽培技术的研究进展[J]. 中国农学通报,2014,30(18):8-13.
[7]杨丽. 垄膜沟播垄沟宽对胡麻水分利用效率和产量的影响[J]. 灌溉排水学报,2019,38(10):17-25.
[8]李小燕,张雷,牛芬菊,等. 旱地组合型微垄全膜不同覆盖时期对土壤水分及胡麻生长的影响[J]. 干旱地区农业研究,2015,33(2):16-21.
[9]JEONG J H, RESOP J P, MUELLER N D, et al. Random forests for global and regional crop yield predictions[J]. PLoS One,2016,11(6):e0156571.
[10]严海军,卓越,李茂娜,等. 基于机器学习和无人机多光谱遥感的苜蓿产量预测[J]. 农业工程学报,2022,38(11):64-71.
[11]丛佳慧,田兴帅,赵向阳,等. 基于DSSAT模型和天气预报策略预测农户当季玉米产量[J]. 玉米科学,2022,30(4):62-72.
[12]郑昌玲,张蕾,侯英雨,等. 基于WOFOST模型的冬小麦产量动态预报方法[J]. 干旱地区农业研究,2022,40(6):242-250,267.
[13]RUGIMBANA C. Predicting maize (Zea mays) yields in eastern province of Rwanda using aquacrop model[D]. Nairobi:University of Nairobi,2019.
[14]SARKER I H. Deep learning:a comprehensive overview on techniques,taxonomy,applications and research directions[J]. SN computer science,2021,2(6):1-20.
[15]KHAKI S, WANG L Z, ARCHONTOULIS S V. A CNN-RNN framework for crop yield prediction[J]. Frontiers in Plant Science,2020,10:1750.
[16]于珍珍,邹华芬,于德水,等. 融合田间水热因子的甘蔗产量GA-BP预测模型[J]. 农业机械学报,2022,53(10):277-283.
[17]崔小茹,陈其鲜. 甘肃省胡麻生产现状及发展思路[J]. 甘肃农业,2014(11):3-4.
[18]栾振斌,宁和平,姚瑞,等. 1965—2015年甘肃省干旱时空分布特征分析[J]. 现代农业科技,2022(13):124-129.
[19]LI M Y, CAO S, ZHU Z C, et al. Spatiotemporally consistent global dataset of the GIMMS Normalized Difference Vegetation Index (PKU GIMMS NDVI) from 1982 to 2022[J]. Earth System Science Data,2023,15(9):4181-4203.
[20]郝洪涛,王凯,张炳建,等. 多尺度特征自适应融合的气动控制阀故障诊断[J]. 仪器仪表学报,2023,44(10):167-178.
[21]李笛,杨东,王文庆,等. 基于CNN-LSTM-Attention的工业控制系统网络入侵检测方法研究[J]. 热力发电,2024,53(5):115-121.
[22]兰小机,贺永兰,武帅文. 基于RF-BiLSTM模型的河流水质预测[J]. 长江科学院院报,2024,41(7):57-63,71.
[23]GRAVES A, SCHMIDHUBER J. Framewise phoneme classification with bidirectional LSTM and other neural network architectures[J]. Neural Networks,2005,18(5/6):602-610.
[24]雷冰冰,韩镏,石佳圆,等. 基于CRITIC和多策略秃鹰优化BiLSTM的水质预测研究[J]. 安全与环境学报,2024,24(9):3688-3702.
[25]BAHDANAU D, CHO K, BENGIO Y. Neural machine translation by jointly learning to align and translate[J/OL]. arXiv,2014.
[2024-10-03].https://doi.org/10.48550/arXiv.1409.0473.DOI:10.48550/arXiv.1409.0473.
[26] HU J, SHEN L, SUN G. Squeeze-and-excitation networks[C]//IEEE,CVF. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Salt Lake City:IEEE,2018.
[27]吴瑞香,杨建春. 主要气象因子与胡麻产量性状的相关性分析[J]. 耕作与栽培,2023,43(6):6-10.

相似文献/References:

[1]樊玉春,李玥,魏霖静,等.三种胡麻生长模型对现蕾期和青果期干旱胁迫响应能力的比较[J].江苏农业学报,2023,(02):423.[doi:doi:10.3969/j.issn.1000-4440.2023.02.015]
　FAN Yu-chun,LI Yue,WEI Lin-jing,et al.Comparison of responsiveness of three oil flax growth models to drought stress at budding stage and green fruit stage[J].,2023,(07):423.[doi:doi:10.3969/j.issn.1000-4440.2023.02.015]

备注/Memo

备注/Memo:
收稿日期:2024-11-19基金项目:国家自然科学基金项目(32460443、32060437)；甘肃省科技计划-自然科学基金重点项目(23JRRA1403)作者简介:李星宇(1998-),男,河南信阳人,硕士研究生,研究方向为作物产量预测。(E-mail)1626850411@qq.com通讯作者:李玥,(E-mail)liyue@gsau.edu.cn
更新日期/Last Update: 2025-08-19