位置:成果数据库 > 期刊 > 期刊详情页
煤尘润湿接触角的BP神经网络估算方法研究
  • ISSN号:1009-6094
  • 期刊名称:《安全与环境学报》
  • 时间:0
  • 分类:X936[环境科学与工程—安全科学]
  • 作者机构:[1]辽宁工程技术大学环境科学与工程学院,辽宁阜新123000, [2]大连交通大学土木与安全工程学院,辽宁大连116028
  • 相关基金:国家自然科学基金项目(U1261121)
中文摘要:

通过采集和测定35个矿区煤样品的化学组成、结构参数和润湿接触角,构建了以13个影响因子为输入参数和以接触角为输出目标的3层BP人工神经网络,并利用该模型估算煤尘润湿接触角。结果表明,隐含层节点数为19时,接触角估算值与实测值的决定系数R2=0.957,平均相对误差为4.59%,表明基于BP神经网络建立的煤尘润湿接触角估算模型具有很高的精度。

英文摘要:

The article intends to present a new approach to the estimation of the contact angle of the coal dust moistening based on BP neural network. For the study purpose,we have collected working face coal samples from the 35 various coal mining areas,and then examined and measured the coal quality parameters in the laboratory condition. What is more,we have done elemental analysis and industrial analysis over the coal samples in accord-ance with the GB/T 476—2008 and GB/T 212—2008 Standards individually,and measured the specific pore volumes and the average hole diameters of the coal samples with an automatic nitrogen sorption analyzer. At the same time,the moistening contact angles can also be measured with a contact angle analyzer. In addition,the relevant programs can be prepared by using a MATLAB software,while the 13 factors to be input as the key parameters can be used to result in a three layer BP artificial neural network with two logarithmic Sigmoid type functions. And,finally,the established model can be applied to estimating the coal dust moistening contact angle. The results of our investigation and examination show that:( 1) We have found that there exists an obvious non-linear relationship between the moistening contact angles and the quality parameters of the coal samples we have collected from the various coal mining areas;( 2) In the BP neural network training, with the increasing number of hidden layer nodes,the mean output errors and the standard deviations may decrease first and then begin to increase;( 3) The number of hid-den layer nodes tend to have great effects on the relative estimation errors of the coal samples,whereas the proper number of hidden layer nodes is potential to be able to improve the generalizing performance and identification of the neural network,which are conducive to reducing the relative errors;( 4) The determination coefficient between the measured values and the estimated values of the coal dust moistening contact angle has been found equal to0. 957

同期刊论文项目
同项目期刊论文
期刊信息
  • 《安全与环境学报》
  • 北大核心期刊(2011版)
  • 主管单位:中国兵器工业集团公司
  • 主办单位:北京理工大学 中国环境科学学会 中国职业安全健康协会
  • 主编:冯长根
  • 地址:北京市海淀区中关村南大街5号
  • 邮编:100081
  • 邮箱:aqyhjxb@263.net;aqyhjxb@wuma.com.cn
  • 电话:010-68913997
  • 国际标准刊号:ISSN:1009-6094
  • 国内统一刊号:ISSN:11-4537/X
  • 邮发代号:2-770
  • 获奖情况:
  • 获首届《CAJ-CD》执行优秀期刊奖,中国科技论文统计源期刊
  • 国内外数据库收录:
  • 美国化学文摘(网络版),中国中国科技核心期刊,中国北大核心期刊(2008版),中国北大核心期刊(2011版),中国北大核心期刊(2014版)
  • 被引量:17182