基于BILSTM的棉花价格预测建模与分析

doi:10.13733/j.jcam.issn.2095-5553.2021.08.21

摘要/Abstract

摘要： 棉花市场价格指数波动是一个非常复杂的非线性系统,具有随机波动特性,容易受到气象、金融、政策和国际环境影响。在现有研究棉花价格的数据集特征的基础上如政策、国际环境、进出口、产量等,增加气候因素对棉花价格影响的数据特征如降水、日照、湿度等,并对数据进行收集、整理及预处理。基于棉花价格的波动特性,采用双向长短期记忆网络BiLSTM(bidirectional long short-term memory, BiLSTM)模型对棉花价格进行预测,使用长短期记忆网络LSTM(Long Short Term Memory Network, LSTM)和LightGbm模型进行对比试验。由于随机梯度下降(SGD)优化器在训练的过程中产生频繁波动,较多情况下得到的是局部最优值。采用SWA(Stochastic Weight Averaging)优化算法取SGD轨迹的多点简单平均值对SGD进行优化,避免SGD在梯度下降过程中的频繁波动问题,使其模型能将Loss和损失值收敛至全局最优,进一步提高训练的稳定性。试验结果表明:BILSTM模型能够很好地对测试集价格曲线进行拟合,误差值最小,价格预测精度较高;采用SWA算法优化的LSTM和BILSTM网络结构收敛至全局最优,平均绝对误差(MAE)分别提高18%和43%。该模型能够更精确地表现棉花市场价格波动规律,帮助棉花市场从业者和投资者优化经营策略。

关键词: 神经网络, 长短期记忆网络, 随机权重平均, 棉花价格预测

Abstract: The fluctuation of the cotton market price index is a very complex nonlinear system with random fluctuation characteristics, which is easily affected by weather, finance, policy, and the international environment. In this paper, based on the existing research on the characteristics of the cotton price data set, such as policy, international environment, import and export, production, the data characteristics of the impact of climate factors on cotton prices such as precipitation, sunshine, humidity were added, and the data were analyzed collection, sorting and pretreatment. Based on the volatility characteristics of cotton prices, a bidirectional long shortterm memory (BiLSTM) model to predict cotton prices coupled with Long Short Term Memory Network (LSTM) and LightGbm trees were used in this research. The model was tested for comparison. Since the Stochastic Gradient Descent (SGD) optimizer produced frequent fluctuations during the training process, in most cases, the local optimal value was obtained. This experiment used the SWA (Stochastic Weight Averaging) optimization algorithm to take the multipoint simple average of the SGD trajectory to optimize SGD. This avoided the frequent fluctuation of SGD in the gradient descent process, and enabled the model to converge Loss and loss values to the global optimal to further improve the stability of training. The experimental results showed that the BILSTM model can fit the price curve of the test set well, with the smallest error value and high price prediction accuracy. The LSTM and BILSTM network structures optimized by the SWA algorithm converged to the approximate global optimum, with the average absolute error (MAE) increased by 18% and 43% respectively. This model can represent the law of cotton market price fluctuations more accurately and help cotton market practitioners and investors optimize their business strategies.

Key words: neural network, longshortterm memory network, stochastic weight averaging, cotton price forecast

中图分类号:

F323.7

江知航, 王艳霞, 颜家均, 周堂容, . 基于BILSTM的棉花价格预测建模与分析[J]. 中国农机化学报, 2021, 42(8): 151-160.

Jiang Zhihang, Wang Yanxia, Yan Jiajun, Zhou Tangrong.. Modeling and analysis of cotton price forecast based on BILSTM[J]. Journal of Chinese Agricultural Mechanization, 2021, 42(8): 151-160.

参考文献

［１］　
陆江林, 石磊, 张玉同. 我国棉花总产量波动原因分析［J］. 中国农机化学报, 2016, 37(8): 248-251.

Lu Jianglin, Shi Lei, Zhang Yutong. Analyze the fluctuations of total cotton production in China ［J］. Journal of Chinese Agricultural Mechanization, 2016, 37(8): 248-251.

［2］　
Fayek H M, Lech M, Cavedon L. Evaluating deep learning architectures for speech emotion recognition ［J］. Neural Networks, 2017, 92(2): 60-68.

［3］　
MR CostaJussà, Allauzen A, Barrault L, et al. Introduction to the special issue on deep learning approaches for machine translation ［J］. Computer Speech & Language, 2017, 46(11): 367-373.

［4］　
Xing J, Li K, Hu W, et al. Diagnosing deep learning models for high accuracy age estimation from a single image ［J］. Pattern Recognition, 2017, 66: 106-116.

［5］　
刘斌, 何进荣, 李远成, 等. 基于分布式神经网络的苹果价格预测方法［J］. 计算机应用, 2020, 40(2): 369-374.

Liu Bin, He Jinrong, Li Yuancheng, et al. Apple price prediction method based on distributed neural network ［J］. Journal of Computer Applications, 2020, 40(2): 369-374.

［6］　
吴叶, 刘婷婷, 方少勇. 基于MIVGABP神经网络的我国棉价预测研究［J］. 棉纺织技术, 2018, 46(7): 77-80.

Wu Ye, Liu Tingting, Fang Shaoyong. Forecast of CN cotton B based on MIVGABP neural network ［J］. Cotton Textile Technology, 2018, 46(7): 77-80.

［7］　
郭超. 农产品价格数据挖掘与趋势预测模型的研究［D］. 济南: 山东大学, 2009.

Guo Chao. Farm price data mining and tendency forecast model research ［D］. Jinan: Shandong University, 2009.

［8］　
闫庆华, 刘维忠, 秦子. 国际棉花现货价格波动及短期预测研究［J］. 资源开发与市场, 2017, 33(5): 575-578, 640.

Yan Qinghua, Liu Weizhong, Qin Zi. Study on fluctuation of international cotton spot price and short term prediction ［J］. Resource Development & Market, 2017, 33(5): 575-578, 640.

［9］　
赵梅, 刘维忠. ARIMA和平滑ARIMA模型的中国棉花价格短期预测比较［J］. 贵州农业科学, 2015, 43(11): 206-208.

Zhao Mei, Liu Weizhong. Comparison of shortterm cotton price forecasted by ARIMA model and smooth ARIMA model in China ［J］. Guizhou Agricultural Sciences, 2015, 43(11): 206-208.

［10］　
Hudson D, K Coble. Harvest contract price volatility for cotton ［J］. Journal of Futures Markets, 1999, 19(6): 717-733.

［11］　
Bailey D V, Brorsen B W, Richardson J W. Dynamic stochastic simulation of daily cash and futures cotton prices ［J］. Southern Journal of Agricultural Economics, 1984, 16(2): 109-116.

［12］　
俞书傲. 气候变化对农作物生产的影响——以浙江为例的实证研究［D］. 杭州: 浙江大学, 2019.

Yu Shuao. The impact of climate change on crop production: An empirical study in Zhejiang, China ［D］. Hangzhou: Zhejiang University, 2019.

［13］　
解伟, 魏玮, 崔琦. 气候变化对中国主要粮食作物单产影响的文献计量Meta分析［J］. 中国人口·资源与环境, 2019, 29(1): 82-88.

Xie Wei, Wei Wei, Cui Qi. The impacts of climate change on the yield of staple crops in China: A Metaanalysis ［J］. China Population Resources and Environment, 2019, 29(1): 82-88.

［14］　
王亚飞, 廖顺宝. 气候变化对粮食产量影响的研究方法综述［J］. 中国农业资源与区划, 2018, 39(12): 59-68.

Wang Yafei, Liao Shunbao. Impacts of climate change on grain yield: A review of research methods ［J］. Chinese Journal of Agricultural Resources and Regional Planning, 2018, 39(12): 59-68.

［15］　
赵东东, 赵雅丽, 赵秉强, 等. 基于PCALSTM小麦叶片水分检测研究［J］. 中国农机化学报, 2019, 40(3): 154-158, 184.

Zhao Dongdong, Zhao Yali, Zhao Binqiang, et al. Study on wheat leaf moisture detection based on PCALSTM ［J］. Journal of Chinese Agricultural Mechanization, 2019, 40(3): 154-158, 184.

［16］　
Graves A, Schmidhuber J. Framewise phoneme classification with bidirectional LSTM and other neural network architectures ［J］. Neural Networks, 2005, 18(56): 602-610.

［17］　
彭红星, 郑楷航, 黄国彬, 等. 基于BP、LSTM和ARIMA模型的蔬菜价格预测［J］. 中国农机化学报, 2020, 41(4): 193-199.

Peng Hongxing, Zheng Kaihang, Huang Guobin, et al. Vegetable price prediction based on BP, LSTM and ARIMA models ［J］. Journal of Chinese Agricultural Mechanization, 2020, 41(4): 193-199.

［18］　
滕金玲, 柳平增, 张艳, 等. 基于Prophet的生姜价格预测研究［J］. 中国农机化学报, 2020, 41(8): 211-216.

Teng Jinling, Liu Pingzeng, Zhang Yan, et al. Research on ginger price forecast based on Prophet ［J］. Journal of Chinese Agricultural Mechanization, 2020, 41(8): 211-216.

[1]	曹金凤, 沈大港, 郭继鸿, 刘鹏, 李策, 兰添贺. 基于CNN和SVM分类优化的大蒜鳞芽朝向识别研究[J]. 中国农机化学报, 2022, 43(5): 134-139.
[2]	雷雪梅, 张光强, 姚旗, 刘伟渭, 邱帅. 基于卷积神经网络的农机图像自动识别研究[J]. 中国农机化学报, 2022, 43(5): 140-147.
[3]	孙博瑞, , 孙三民, , 蒋敏, , 薛山. 基于LSTM神经网络的智能灌溉系统设计与试验#br# #br# [J]. 中国农机化学报, 2022, 43(4): 116-123.
[4]	蒋柏春, 李德仑, 韦克苏, 张富贵, 王杰, 刘红芸. 基于高光谱的烤烟叶绿素含量估算模型研究[J]. 中国农机化学报, 2022, 43(3): 104-110.
[5]	高文强, , 肖志云, . 多尺度特征融合1D—CNN的马铃薯植株高光谱数据地物分类和缺素识别#br#[J]. 中国农机化学报, 2022, 43(3): 111-119.
[6]	罗兵, , 张建敏. 基于BP神经网络的农机安全评价敏感性分析[J]. 中国农机化学报, 2022, 43(3): 120-126.
[7]	董萍, 卫梦华, 时雷, 郭伟. 迁移学习在玉米叶片病害识别中的研究与应用[J]. 中国农机化学报, 2022, 43(3): 146-152.
[8]	孙浩楠, 仝志民, 谢秋菊, 李嘉熙. 基于BP神经网络猪咳嗽声识别[J]. 中国农机化学报, 2022, 43(2): 148-154.
[9]	石洪康, 肖文福, 黄亮, 胡丛武, 胡光荣, 张剑飞. 基于卷积神经网络的家蚕病害识别研究[J]. 中国农机化学报, 2022, 43(1): 150-.
[10]	宋卓研, 徐晓辉, 宋涛, 崔迎港, 司玉龙, . 基于PSO-BP优化PID模型的水肥控制系统研究[J]. 中国农机化学报, 2021, 42(9): 83-89.
[11]	李小敏, 张日红, 陈天赐, 侯炳法, 张权, 李雄, . 基于深度学习的林下落果识别方法与试验[J]. 中国农机化学报, 2021, 42(9): 202-208.
[12]	王鹤树, 曹丽英, . 基于全卷积神经网络的植物叶片自动分割及表型解析[J]. 中国农机化学报, 2021, 42(8): 161-168.
[13]	张建超, 单慧勇, 景向阳, 李晨阳, 张程皓, 刘慧芹, . 基于Elman神经网络的温室环境因子预测方法[J]. 中国农机化学报, 2021, 42(8): 203-208.
[14]	刘海涛;伊丽丽;兰玉彬;韩鑫;崔立华;. 机器视觉在棉花智能打顶领域的应用研究进展[J]. 中国农机化学报, 2021, 42(6): 159-165.
[15]	陈强;戴军;. 农机驾驶人员疲劳驾驶行为检测研究[J]. 中国农机化学报, 2021, 42(5): 148-152.