无机盐工业
主管:中海油天津化工研究设计院有限公司
主办:中海油天津化工研究设计院有限公司
   中海油炼油化工科学研究院(北京)有限公司
   中国化工学会无机酸碱盐专业委员会
ISSN 1006-4990 CN 12-1069/TQ

无机盐工业 ›› 2024, Vol. 56 ›› Issue (1): 53-58.doi: 10.19964/j.issn.1006-4990.2023-0078

• 研究与开发 • 上一篇    下一篇

基于反向传播神经网络的卤水蒸发速率预测模型

李志伟1,2,3(), 付振海1,2,3(), 张志宏1,2,3, 李生廷4   

  1. 1.中国科学院青海盐湖研究所,青海 西宁 810008
    2.中国科学院盐湖资源综合高效利用重点实验室,青海 西宁 810008
    3.青海省盐湖资源化学重点实验室,青海 西宁 810008
    4.青海盐湖工业股份有限公司,青海格尔木 816000
  • 收稿日期:2023-02-16 出版日期:2024-01-10 发布日期:2024-01-18
  • 通讯作者: 付振海(1985— ),男,副研究员,主要研究方向为盐湖资源综合利用和荧光探针;E-mail:fzh@isl.ac.cn
  • 作者简介:李志伟(1996— ),男,硕士,主要研究方向为盐湖资源综合利用;E-mail:lizw@isl.ac.cn
  • 基金资助:
    青海省科技厅项目(2021-GX-102);中国科学院STS区域重点项目(KFJ-STS-QYZD2021-06-001)

Prediction model of brine evaporation rate based on back-propagation neural network

LI Zhiwei1,2,3(), FU Zhenhai1,2,3(), ZHANG Zhihong1,2,3, LI Shengting4   

  1. 1. Key Laboratory of Comprehensive and Highly Efficient Utilization of Salt Lake Resources,Xining,810008,China
    2. Qinghai Institute of Salt Lake,Chinese Academy of Sciences,Xining 810008,China
    3. Key Laboratory of Salt Lake Resources Chemistry of Qinghai Province,Xining 810008,China
    4. Qinghai Salt Lake Industry Co.,Ltd. of Qinghai Province,Golmud 816000,China
  • Received:2023-02-16 Published:2024-01-10 Online:2024-01-18

摘要:

卤水的蒸发速率是盐田生产管理中的一个重要技术参数,通过搭建室外卤水蒸发实验装置,分析了辐照强度、风速、环境温度、相对湿度、卤水温度、卤水浓度与卤水蒸发速率的关系。利用反向传播(BP)神经网络,训练构建了卤水蒸发速率预测模型,并与传统的应用回归方法构建的模型进行比较。结果表明,BP神经网络模型和非线性回归模型的决定系数R2分别为0.902和0.884,预测平均相对误差分别为15.723%和18.943%,BP神经网络模型的拟合效果和预测能力均优于非线性回归模型。说明应用BP神经网络构建卤水蒸发速率预测模型是可行的,能够实现蒸发速率的快速估测。

关键词: 卤水蒸发速率, 定量分析, 非线性回归, 反向传播神经网络

Abstract:

Brine evaporation rate is an important technical parameter in the production and management of salt pans.By setting up an outdoor brine evaporation experimental device,the relationship between irradiation intensity,wind speed,ambient temperature,relative humidity,brine temperature,brine concentration,and brine evaporation rate was analyzed.The prediction model of brine evaporation rate was constructed by using back-propagation(BP) neural network and compared with the model constructed by traditional regression method.The results showed that the determination coefficients R2 of BP neural network model and nonlinear regression model were 0.902 and 0.884,respectively,and the average relative error were 15.723% and 18.943%,respectively.It was indicated that the fitting effect and prediction ability of BP neural network model were better than nonlinear regression model.It was feasible to use BP neural network to construct the prediction model of brine evaporation rate,which could realize the rapid estimation of evaporation rate.

Key words: brine evaporation rate, quantitative analysis, nonlinear regression, back-propagation neural network

中图分类号: