基于LSTM神经网络的苹果价格预测模型  

Price forecasting model of apple based on LSTM neural net work

作　　者：张顺利[1] 王小东 李艳翠[1] ZHANG Shunli;WANG Xiaodong;LI Yancui(School of Information Engineering,Henan Institute of Science and Technology,Xinxiang 453003,China;Shandong Liaocheng No 2 Middle School,Liaocheng 252000,China)

机构地区：[1]河南科技学院信息工程学院,河南新乡453003 [2]山东聊城第二中学,山东聊城252000

出　　处：《河南科技学院学报（自然科学版）》2020年第5期73-78,共6页Journal of Henan Institute of Science and Technology(Natural Science Edition)

基　　金：国家自然科学基金(61502149);河南省重点科研研究项目(15A520069);河南省科技厅科技攻关(182102210048)。

摘　　要：针对苹果价格数据随时间变化的非平稳性、季节性和周期性特征,提出一种基于长短期记忆网络(Long-short Term Memory,LSTM)的神经网络价格预测模型.选用苹果的历史价格、可替补水果的历史价格和居民生活消费指数为特征进行模型的输入,然后使用BLSTM网络学习苹果的价格表示,最后输出苹果的预测价格.选用豫南阳果品批发交易中心的烟台红富士价格数据进行实验,结果表明:使用基于LSTM神经网络预测模型的平均绝对误差、均方误差和均方根误差均小于传统的差分整合移动平均自回归(Autoregressive Integrated Moving Average Model,ARIMA)模型.Aimed at the non-stationary,seasonal and cyclical features of appleprice time series,the model based on LSTM(long-short term memory)was proposed to forecast the future prices of apple.The history prices of apple and substitution fruits of apple,the level of household consumption were selected as the input of the neural network,then the BLSTM layer was utilized to model apple prices,finally the forecasting prices of apples were shown.The market prices of apples in Henan Nanyangfruits trade centerwere selected as experimental data.Compared with thepreviousautoregressive integrated moving average(ARIMA)model,the proposed LSTM model can reduce the error in MAE,MSE and RMSE.

关 键 词：苹果 价格预测 LSTM 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]