基于免疫原理的径向基神经网络品位插值研究
|
摘要
针对地质统计学方法的应用缺陷,基于径向基函数神经网络(RBF网络),结合免疫算法,开展矿石品位插值研究。利用RBF网络对样本数据进行分类及训练,通过免疫算法进行数据聚类分析,确定RBF网络的隐含层节点数、径向基函数中心向量及其宽度等参数。在此基础上,选取某典型矿山品位数据进行插值计算,将插值结果与品位真实值及克立格插值进行比较分析。研究结果表明:所给出的插值模型计算效率高,算法可以覆盖更多的训练数据,全局寻优能力强;插值结果具有较高的精度。当矿床无法满足地质统计学使用条件时,可考虑采用神经网络方法对矿石品位进行估值计算。
For the application defects of geostatistical methods,the study of ore grade interpolation based on radial basis function neural network(RBF network)and combined with immune algorithm was carried out.The RBF network was used to classify and train the sample data,and the data clustering analysis was performed through the immune algorithm to determine the number of hidden layer nodes,the center vector and width of the radial basis function.On this basis,the grade data of a typical mine was selected for interpolation calculation and the interpolation results were compared with the real grade value and Kriging interpolation.The results showed that the interpolation model presented in this paper was efficient,the algorithm could cover more training data,and the ability of global optimization was strong.The interpolation result was of high precision.When the ore deposit cannot meet the use conditions of geological statistics,the neural network method could be used to evaluate the ore grade.
For the application defects of geostatistical methods,the study of ore grade interpolation based on radial basis function neural network(RBF network)and combined with immune algorithm was carried out.The RBF network was used to classify and train the sample data,and the data clustering analysis was performed through the immune algorithm to determine the number of hidden layer nodes,the center vector and width of the radial basis function.On this basis,the grade data of a typical mine was selected for interpolation calculation and the interpolation results were compared with the real grade value and Kriging interpolation.The results showed that the interpolation model presented in this paper was efficient,the algorithm could cover more training data,and the ability of global optimization was strong.The interpolation result was of high precision.When the ore deposit cannot meet the use conditions of geological statistics,the neural network method could be used to evaluate the ore grade.
引文
[1]Emery X.Uncertainty modeling and spatial prediction by multi-Gaussian Kriging:Accounting for an unknown mean value[J].Computers&Geosciences,2008,34(11):1431-1442.
[2]刘修国,尤超,张唯.非线性克里格法在矿体储量估算中的应用[J].中南大学学报(自然科学版),2013,44(5):2068-2072.
[3]金毅,郑德顺,潘懋,等.各向异性矿化特征的分析与模拟及其在储量估算中的应用[J].地质学报,2011,85(9):1519-1527.
[4]谭正华,荆永滨,王李管,等.基于空间变异性的IDW矿石品位估值改进方法[J].中国矿业大学学报,2011,40(6):928-932.
[5]李翠平,李仲学,余东明.基于泰森多边形法的空间品位插值[J].辽宁工程技术大学学报,2007,26(4):488-491.
[6]孙卡,翁正平,李章林,等.排序扫描线算法在克里格估值中的应用[J].金属矿山,2009,(7):73-76.
[7]郭奇峰,刘金辉.析取克里格法在可回采资源量的总体估算的应用[J].中国矿业,2012,21(8):20-22.
[8]孙刚,徐兵,韩燕.限制性克立格法在地学中的研究与应用[J].地球物理学进展,2008,23(2):617-621.
[9]Clarici E,Owen D,Durucan S,et al.Recoverable reserve estimation using a neural network[A].Proceedings of the24th International Symposium on the Application of Computers and Operations Research in the Minerals Industries(APCOM)[C]//Montreal,1993:145-152.
[10]Wu X,Zhou Y.Reserve estimation using neural network techniques[J].Computers&Geosciences,1993,19(4):567-575.
[11]Burnett C C H.Application of neural networks to mineral reserve estimation[D].Nottingham:University of Nottingham,1995.
[12]Cortez L P,Sousa A J,Durao F O.Mineral resources estimation using neural networks and geostatistical techniques[A].27th Internation Symposium on Computer Applications in the Mineral Industries(APCOM 98)[C]//London,1998:305-314.
[13]Kapageridis I K.Application of artificial neural network systems to grade estimation from exploration data[D].Nottingham:University of Nottingham,1999.
[14]韩万林,张幼蒂.用改进BP算法估算矿石品位[J].中国矿业,2000,9(3):80-82.
[15]卢才武,云庆夏.矿块品位计算的神经网络方法[J].金属矿山,2003,(6):41-43.
[16]董卫军.矿床品位估值网络泛化能力研究[J].中国矿业,1999,8(5):118-120.
[17]周仲礼,马腾,陈秀荣,等.基于改进RBF的空间插值算法及其在矿体三维可视化中的应用[J].成都理工大学学报(自然科学版),2014,41(5):645-650.
[18]贾明涛,叶加冕,寇向宇,等.品位估值的自适应径向基神经网络构建技术[J].煤炭学报,2010,35(9):1524-1530.
[19]吴永忠,程文娟,韩江洪,等.一种新型的红外甲烷测量定量分析模型[J].煤炭学报,2009,34(2):275-279.
[20]Rawlings J O.Applied regression analysis[M].Pacific Grove:Wadsworth&Brooks/Cole,1988.
[21]李眩.基于免疫原理的矿石品位估算[D].西安:西安建筑科技大学,2008.
[2]刘修国,尤超,张唯.非线性克里格法在矿体储量估算中的应用[J].中南大学学报(自然科学版),2013,44(5):2068-2072.
[3]金毅,郑德顺,潘懋,等.各向异性矿化特征的分析与模拟及其在储量估算中的应用[J].地质学报,2011,85(9):1519-1527.
[4]谭正华,荆永滨,王李管,等.基于空间变异性的IDW矿石品位估值改进方法[J].中国矿业大学学报,2011,40(6):928-932.
[5]李翠平,李仲学,余东明.基于泰森多边形法的空间品位插值[J].辽宁工程技术大学学报,2007,26(4):488-491.
[6]孙卡,翁正平,李章林,等.排序扫描线算法在克里格估值中的应用[J].金属矿山,2009,(7):73-76.
[7]郭奇峰,刘金辉.析取克里格法在可回采资源量的总体估算的应用[J].中国矿业,2012,21(8):20-22.
[8]孙刚,徐兵,韩燕.限制性克立格法在地学中的研究与应用[J].地球物理学进展,2008,23(2):617-621.
[9]Clarici E,Owen D,Durucan S,et al.Recoverable reserve estimation using a neural network[A].Proceedings of the24th International Symposium on the Application of Computers and Operations Research in the Minerals Industries(APCOM)[C]//Montreal,1993:145-152.
[10]Wu X,Zhou Y.Reserve estimation using neural network techniques[J].Computers&Geosciences,1993,19(4):567-575.
[11]Burnett C C H.Application of neural networks to mineral reserve estimation[D].Nottingham:University of Nottingham,1995.
[12]Cortez L P,Sousa A J,Durao F O.Mineral resources estimation using neural networks and geostatistical techniques[A].27th Internation Symposium on Computer Applications in the Mineral Industries(APCOM 98)[C]//London,1998:305-314.
[13]Kapageridis I K.Application of artificial neural network systems to grade estimation from exploration data[D].Nottingham:University of Nottingham,1999.
[14]韩万林,张幼蒂.用改进BP算法估算矿石品位[J].中国矿业,2000,9(3):80-82.
[15]卢才武,云庆夏.矿块品位计算的神经网络方法[J].金属矿山,2003,(6):41-43.
[16]董卫军.矿床品位估值网络泛化能力研究[J].中国矿业,1999,8(5):118-120.
[17]周仲礼,马腾,陈秀荣,等.基于改进RBF的空间插值算法及其在矿体三维可视化中的应用[J].成都理工大学学报(自然科学版),2014,41(5):645-650.
[18]贾明涛,叶加冕,寇向宇,等.品位估值的自适应径向基神经网络构建技术[J].煤炭学报,2010,35(9):1524-1530.
[19]吴永忠,程文娟,韩江洪,等.一种新型的红外甲烷测量定量分析模型[J].煤炭学报,2009,34(2):275-279.
[20]Rawlings J O.Applied regression analysis[M].Pacific Grove:Wadsworth&Brooks/Cole,1988.
[21]李眩.基于免疫原理的矿石品位估算[D].西安:西安建筑科技大学,2008.