LSTM-Prophet混合模型在物料储备需求预测中的应用
Application of LSTM-PROPHET mixed model in material reserve demand forecasting
辛唯1
作者信息
- 1. 青岛科技大学信息科学技术学院,青岛 266000
- 折叠
摘要
建立LSTM-Prophet混合模型,预测A公司未来每日物料消耗量,为工段每日物料储备提供参考,数据集来源为A公司2015年来每日物料实际消耗量.实验结果显示,LSTM-Prophet混合模型预测时序数据的MAE值为40.905,MAPE值为0.044,R2值为0.863,三个评价指标都优于LSTM模型、Prophet模型、ARIMA模型三个单一模型的评价指标值.验证了LSTM-Prophet混合模型有着更好的预测精度,可以更好地应用于工业场景.
Abstract
The LSTM-Prophet mixed model was established to predict the daily material consumption of Company A and pro-vide reference for the daily material reserve of the workshop.The data set was derived from the actual daily material consumption of Company A since 2015.The experimental results showed that the MAE value,MAPE value and R2 value of the LSTM-Prophet mixed model were 40.905,0.044 and 0.863,all of which were better than the LSTM model,Prophet model and ARIMA model.It is verified that the LSTM-Prophet hybrid model has better prediction accuracy and can be better applied to industrial scenarios.
关键词
LSTM模型/Prophet模型/时序数据预测/混合模型预测Key words
LSTM model/Prophet model/time series data prediction/mixed model prediction引用本文复制引用
出版年
2024
NETL
NSTL
维普
万方数据