流体机械

2021, v.49;No.587(05) 87-90+104

[打印本页] [关闭]
本期目录(Current Issue) | 过刊浏览(Past Issue) | 高级检索(Advanced Search)

基于数据挖掘的冷藏陈列柜的负荷预测研究
Research on load forecasting of refrigerated display cabinet based on data mining

袁培;雷正霖;曾庆辉;武宜霄;吕彦力;胡朝龙;
YUAN Pei;LEI Zhenglin;ZENG Qinghui;WU Yixiao;LYU Yanli;HU Chaolong;Zhengzhou University of Light Industry;Shandong duckling cold chain Co.,Ltd.;

摘要(Abstract):

为了挖掘冷藏陈列柜运行数据的有效信息,实现冷藏陈列柜制冷负荷的预测,针对运行数据存在时间序列性和非线性等特性,提出了基于深度学习的LSTM神经网络模型。将日期信息、蒸发器进出口温度、压力、焓值、流量数据按照时间步长划分为训练集与测试集,模型导入后经过训练证明最优方案:模型网络层数为4、训练次数为200,将预测负荷数据与实际负荷进行比对,发现预测模型绝对百分比误差仅为1.98%,具备较好的预测性能。
In order to mine the effective information of the operation data of refrigerated display cabinets and realize the prediction of refrigeration load of refrigerated display cabinets,the LSTM neural network model based on deep learning is proposed for the time series and nonlinear characteristics of the operation data.The date information,evaporator inlet and outlet temperature,pressure,enthalpy value,and flow data are divided into training set and test set according to the time step,after the model is imported and trained,it is proved that the optimal solution is as follows:4 layers of networks of the model and 200 training times.After comparing the forecasted load data with the actual load,it is found that the absolute percentage error of the forecasting model is only 1.98%,which has a better forecasting performance.

关键词(KeyWords): 负荷预测;LSTM神经网络模型;损失函数;冷藏陈列柜
load forecasting;LSTM neural network model;loss function;refrigerated display cabinet

Abstract:

Keywords:

基金项目(Foundation): 国家自然科学基金项目(51476148);; 河南省制冷与低温设备节能技术创新科技团队(CXTD2011042)

作者(Author): 袁培;雷正霖;曾庆辉;武宜霄;吕彦力;胡朝龙;
YUAN Pei;LEI Zhenglin;ZENG Qinghui;WU Yixiao;LYU Yanli;HU Chaolong;Zhengzhou University of Light Industry;Shandong duckling cold chain Co.,Ltd.;

Email:

DOI:

参考文献(References):

扩展功能
本文信息
服务与反馈
本文关键词相关文章
本文作者相关文章
中国知网
分享