[1]勾志竟,宫志宏 *,刘布春.基于 TensorFlow 的 LSTM 算法在农业中的应用[J].计算机技术与发展,2021,31(08):215-220.[doi:10. 3969 / j. issn. 1673-629X. 2021. 08. 037]
　GOU Zhi-jing,GONG Zhi-hong *,LIU Bu-chun.Application of LSTM in Rice Yield Prediction Based on TensorFlow[J].,2021,31(08):215-220.[doi:10. 3969 / j. issn. 1673-629X. 2021. 08. 037]
点击复制

基于 TensorFlow 的 LSTM 算法在农业中的应用()
分享到：

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:
31
期数:
2021年08期
页码:
215-220
栏目:
应用前沿与综合
出版日期:
2021-08-10

文章信息/Info

Title:
Application of LSTM in Rice Yield Prediction Based on TensorFlow
文章编号:
1673-629X(2021)08-0215-06
作者:
勾志竟1; 宫志宏2 *; 刘布春3
1. 天津市气象信息中心,天津 300074;
2. 天津市气候中心,天津 300074;
3. 中国农业科学院农业环境与可持续发展研究所,北京 100081
Author(s):
GOU Zhi-jing1; GONG Zhi-hong2 *; LIU Bu-chun3
1. Tianjin Meteorological Information Center,Tianjin 300074,China;
2. Tianjin Climate Center,Tianjin 300074,China;
3. Institute of Environment and Sustainable Development in Agriculture,Chinese Academy of Agricultural Sciences,Beijing 100081,China
关键词:
水稻产量; BP 神经网络; LSTM; TensorFlow; 预测
Keywords:
rice yield; BP neural network; LSTM; TensorFlow; prediction
分类号:
TP39
DOI:
10. 3969 / j. issn. 1673-629X. 2021. 08. 037
摘要:
针对如何反映水稻产量与生长周期气象要素的非线性关系,并提高产量预测准确度的问题,提出了在 TensorFlow深度学习框架上构建长短时记忆网络算法(LSTM)的水稻产量预测方法。 该方法避免了传统 BP 神经网络容易陷入局部最优和长期预测精度不高的问题,并以天津市宁河区 1989 ~ 2015 年地面气象观测资料与产量数据为基础,选取移栽-返青期、分蘖期、孕穗期、抽穗期、成熟期 5 个不同生长期的风速、日照、温度等 21 个变量作为预测因子,最后以水稻亩产量(kg)作为预测目标进行了实验。 结果表明,BP 神经网络预测结果的均方根误差(root mean square error,RMSE)和平均绝对误差(mean absolute error,MAE)分别为 75. 12 和 65. 64,LSTM 的 RMSE 和 MAE 分别为 34. 77 和 33. 37,相比 BP 神经网络,LSTM 长短期记忆网络的预测精度更高,可以较好地预测水稻产量的长期发展趋势,为水稻生长期的精准管理和决策提供了新的思路和方法。
Abstract:
To solve the problem of how to reflect the nonlinear relationship between rice yield and meteorological factors in the growth period and improve the accuracy of yield prediction,we propose a rice yield forecast method based on the LSTM on TensorFlow deeplearning framework,which can avoid the problem of local optimization and? low accuracy in long - term prediction in traditional BP algorithm. Then,the ground meteorological observation data and the rice statistical data of Ninghe district in Tianjin from the year 1989to 2015 were used,and 21 factors were extracted such as wind speed,sunshine and temperature in 5 different growth stages including transplanting-returning green stage,tillering stage,booting stage,heading stages,maturing stage. Finally,the experiment is performed taking the rice yield per acre as the prediction target. It is showed that the root mean square error (RMSE) and mean absolute error(MAE) of BP neural network were 75. 12 and 65. 64,respectively,the RMSE and MAE of LSTM were 34. 77 and 33. 37. Compared with BP neural network,LSTM has higher prediction accuracy and can better predict the long-term development trend of rice yield,which can be used as a new method for the precise management and decision making in the growth period of rice.

相似文献/References:

[1]浦佳祺,陈德旺.基于最小二乘法和BP神经网络的TOA 定位算法[J].计算机技术与发展,2018,28(05):5.[doi:10.3969/j.issn.1673-629X.2018.05.002]
　PU Jia-qi,CHEN De-wang.A TOA Positioning Algorithm Based on Least Square Method and BP Neural Network[J].,2018,28(08):5.[doi:10.3969/j.issn.1673-629X.2018.05.002]
[2]王嵘冰,徐红艳,李 波,等.BP 神经网络隐含层节点数确定方法研究[J].计算机技术与发展,2018,28(04):31.[doi:10.3969/ j. issn.1673-629X.2018.04.007]
　WANG Rong-bing,XU Hong-yan,LI Bo,et al.Research on Method of Determining Hidden Layer Nodes in BP Neural Network[J].,2018,28(08):31.[doi:10.3969/ j. issn.1673-629X.2018.04.007]
[3]李春生,李霄野,张可佳.基于遗传算法改进的 BP 神经网络房价预测分析[J].计算机技术与发展,2018,28(08):144.[doi:10.3969/ j. issn.1673-629X.2018.08.030]
　LI Chun-sheng,LI Xiao-ye,ZHANG Ke-jia.Price Forecasting Analysis of BP Neural Network Based on Improved Genetic Algorithm[J].,2018,28(08):144.[doi:10.3969/ j. issn.1673-629X.2018.08.030]
[4]辛月振,孙贝贝,夏盛瑜.数据挖掘方法在生物实验数据上的应用[J].计算机技术与发展,2018,28(09):143.[doi:10.3969/ j. issn.1673-629X.2018.09.029]
　XIN Yue-zhen,SUN Bei-bei,XIA Sheng-yu.Application of Data Mining Method in Biological Experiment Data[J].,2018,28(08):143.[doi:10.3969/ j. issn.1673-629X.2018.09.029]
[5]丁红卫,王文果,万 良,等.基于 BP 神经网络的电网物资需求预测研究[J].计算机技术与发展,2019,29(06):138.[doi:10. 3969 / j. issn. 1673-629X. 2019. 06. 029]
　DING Hong-wei,WANG Wen-guo,WAN Liang,et al.Research on Material Demand Prediction of Power Network Based on BP Neural Network[J].,2019,29(08):138.[doi:10. 3969 / j. issn. 1673-629X. 2019. 06. 029]
[6]姬晓飞,石宇辰.多分类器融合的光学遥感图像目标识别算法[J].计算机技术与发展,2019,29(11):52.[doi:10. 3969 / j. issn. 1673-629X. 2019. 11. 011]
　JI Xiao-fei,SHI Yu-chen.Optical Remote Sensing Image Object Recognition Based on Multiple Classifications Fusion[J].,2019,29(08):52.[doi:10. 3969 / j. issn. 1673-629X. 2019. 11. 011]
[7]邓晓政,叶 冰.免疫 BP 网络的机载嵌入式训练系统效能评估[J].计算机技术与发展,2019,29(12):173.[doi:10. 3969 / j. issn. 1673-629X. 2019. 12. 031]
　DENG Xiao-zheng,YE Bing.Effectiveness Evaluation of Airborne Embedded Training System of Immune BP Networks[J].,2019,29(08):173.[doi:10. 3969 / j. issn. 1673-629X. 2019. 12. 031]
[8]李佩钰.一种基于小波和神经网络的短时交通流量预测[J].计算机技术与发展,2020,30(01):135.[doi:10. 3969 / j. issn. 1673-629X. 2020. 01. 024]
　LI Pei-yu.Short-term Traffic Flow Prediction Based on Wavelet and Neural Network[J].,2020,30(08):135.[doi:10. 3969 / j. issn. 1673-629X. 2020. 01. 024]
[9]宋 波,辛文贤,冯云霞.基于 BP 神经网络的临床路径优化[J].计算机技术与发展,2020,30(04):156.[doi:10. 3969 / j. issn. 1673-629X. 2020. 04. 030]
　SONG Bo,XIN Wen-xian,FENG Yun-xia.Clinical Path Optimization Based on BP Neural Network[J].,2020,30(08):156.[doi:10. 3969 / j. issn. 1673-629X. 2020. 04. 030]
[10]余肖生,宋 锦,任明霞,等.基于伪度量的案例推理改进算法[J].计算机技术与发展,2020,30(10):69.[doi:10. 3969 / j. issn. 1673-629X. 2020. 10. 013]
　YU Xiao-sheng,SONG Jin,REN Ming-xia,et al.Improved Case Reasoning Algorithm Based on Pseudo-Metric[J].,2020,30(08):69.[doi:10. 3969 / j. issn. 1673-629X. 2020. 10. 013]

更新日期/Last Update: 2021-08-10