首页 | 本学科首页   官方微博 | 高级检索  
     检索      

基于多聚合过程神经元网络的宁东大气污染物浓度预测研究
引用本文:常瑞芬,李风军.基于多聚合过程神经元网络的宁东大气污染物浓度预测研究[J].贵州师范大学学报(自然科学版),2012,30(5):98-102.
作者姓名:常瑞芬  李风军
作者单位:宁夏大学数学与计算机学院,宁夏银川,750021
基金项目:国家自然科学基金项目(61063020)
摘    要:已有的基于人工神经网络对大气环境质量预测的研究大多只考虑空间特性,因此,无法准确刻画大气环境中污染物浓度随时间的动态变化规律,更不能实现影响污染物浓度诸因子在时间维上的有效预测。鉴于此,主要在时间、空间上对多聚合过程神经元网络模型进行多次训练与学习,并将训练好的模型对宁东能源化工基地大气环境中污染物SO2的浓度进行预测。仿真实验表明:多聚合过程神经元网络对于大气环境中SO2浓度具有较好的预测能力。

关 键 词:多聚合过程神经元网络  大气环境质量  SO2浓度  梯度下降算法

Prediction and study of atmospheric pollutant concentration in ningdong based on multiple polymerization process neural network
CHANG Rui-fen,LI Feng-jun.Prediction and study of atmospheric pollutant concentration in ningdong based on multiple polymerization process neural network[J].Journal of Guizhou Normal University(Natural Sciences),2012,30(5):98-102.
Authors:CHANG Rui-fen  LI Feng-jun
Institution:(School of Mathematics and Computer Science,Ningxia University,Yinchuan,Ningxia 750021,China)
Abstract:In recent years,artificial neural network have been only proposed for solving spatial case of the prediction of atmospheric pollutant concentration,which could not been accurately depicted dynamic change regulation of the atmospheric pollutant concentration in time-varying,So it was not desirable to forecast pollutant concentration in temporal.The purpose of this paper is to present multiple polymerization process neural network model,which based on temporal and spatial scale.By training and learning repeatedly,the trained network model is derived.Then the model is used to forecast SO2 concentration of atmospheric environment in Ningdong Energy and Chemical Base.Simulation results of SO2 concentration demonstrate that the multi-aggregation process neural network is applicability and effectiveness to predict the concentration of the atmospheric environment.
Keywords:multiple polymerization process neural network  atmospheric environment SO2 concentration  gradient descent algorithm
本文献已被 CNKI 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号