基于PCA优化—BP神经网络算法的农田土壤养分评价
|
摘要
为达到对膜下滴灌土壤养分综合评价的目的,采用描述性统计分析、主成分分析(Principal Component Analysis,PCA)和Back Propagation(BP)神经网络方法对通辽市科左中旗农田土壤养分进行评价,结合单因子指数法,并与传统方法中运用最广的模糊数学综合评价结果进行对比分析。依据《全国第二次土壤养分分级标准》,经实测数据验证,结果表明:2015年和2016年科左中旗农田土壤养分综合评价良好;将PCA与BP神经网络相结合的方法运用到解决土壤学问题中,可获得客观的结果;经检验PCA与BP神经网络模型相结合的评价方法适用于因子多且复杂、精度要求高的样本。将PCA与BP神经网络相结合既可以解决信息过多、分析困难的问题,又可以降低模糊性、减少精度下降,将各自优点有效的联系在一起,相较传统方法可以更加科学合理、准确有效地对土壤养分进行综合评价。
In order to achieve the comprehensive evaluation of soil nutrient under drip irrigation,by using descriptive statistical analysis,Principal Component Analysis( PCA) and BP neural network,the soil nutrient with drip irrigation under mulch was evaluated in Kezuo Zhong county of Tongliao city. In combination with the single factor index method,the evaluation result was compared with that of the traditional most widely used fuzzy comprehensive evaluation method. According to the second national soil nutrient grading standard and verification with measured data,the results of this study showed that the comprehensive evaluation of soil nutrients in the farmland of kezuo zhong county was good in 2015 and 2016; more objective results could be obtained when the method of combining PCA and BP neural network was applied to solve soil science problems; the evaluation method combining PCA and BP neural network model was suitable for samples with multiple factors,complex and high precision requirements; the combination of PCA and BP neural network could not only solve the problem of too much information and difficult analysis,but also reduce the fuzziness and accuracy,and effectively link their advantages together. Compared with traditional methods,it is more scientific,reasonable,accurate and effective to comprehensively evaluate soil nutrients.
In order to achieve the comprehensive evaluation of soil nutrient under drip irrigation,by using descriptive statistical analysis,Principal Component Analysis( PCA) and BP neural network,the soil nutrient with drip irrigation under mulch was evaluated in Kezuo Zhong county of Tongliao city. In combination with the single factor index method,the evaluation result was compared with that of the traditional most widely used fuzzy comprehensive evaluation method. According to the second national soil nutrient grading standard and verification with measured data,the results of this study showed that the comprehensive evaluation of soil nutrients in the farmland of kezuo zhong county was good in 2015 and 2016; more objective results could be obtained when the method of combining PCA and BP neural network was applied to solve soil science problems; the evaluation method combining PCA and BP neural network model was suitable for samples with multiple factors,complex and high precision requirements; the combination of PCA and BP neural network could not only solve the problem of too much information and difficult analysis,but also reduce the fuzziness and accuracy,and effectively link their advantages together. Compared with traditional methods,it is more scientific,reasonable,accurate and effective to comprehensively evaluate soil nutrients.
引文
[1] Dumanski J.Land quality indicators:research plan[J]. Transactions of the Chinese Society of Agricultural Engineering,2000,81(2):93-102.
[2]李娟,叶胜兰,牛岩.不同模式对盐碱地土壤环境改良效应的研究[J].中国农村水利水电,2018(9):193-198.
[3] Arnebrant K.Changes in microfungal community structure after fertilization of Scots pine forest soil with ammonium nitrate or urea[J].Transactions of the Chinese Society of Agricultural Engineering,1990,22(3):309-312.
[4] Singh J S,Raghubanshi A S,Singh R S,et al. Microbial biomass acts as a source of plant nutrients in dry tropical forest and savanna[J]. Nature,1989(338):499-500.
[5]马祥华,焦菊英.黄土丘陵沟壑区退耕地自然恢复植被特征及其与土壤环境的关系[J].中国水土保持科学,2005,3(2):15-22.
[6]叶存旺,翟巧绒,郭梓娟,等.沙棘-侧柏混交林土壤养分、微生物与酶活性的研究[J].西北林学院学报,2007,22(5):1-6.
[7]侯琳,雷瑞德,王得祥.黄龙山林区封育油松林土壤养分研究[J].西北农林科技大学学报(自然科学版),2007,35(2):63-68.
[8]韩万海,王增丽.不同灌溉模式对土壤水盐分布及棉花产量的影响[J].中国农村水利水电,2018(11):26-29,34.
[9]柴慧祥.内蒙古通辽市滴灌条件下农田水土环境监测评估[D].呼和浩特:内蒙古农业大学,2016.
[10]吴军斌,尹娟,苏振娟,等.滴灌施肥条件下土壤水分运移规律的研究[J].节水灌溉,2018(11):51-55.
[11]田兆顺,董汉章.华北平原瓦碱的特性和形成[J].土壤学报,1965(1):24-38.
[12] Webster R. Quantitative and numerical methods in soil classification and survey[M]. Clarendon Press,1977:1-7.
[13]赵青,王子龙,姜秋香.属性识别理论模型在土壤肥力综合评价中的应用[J].农业系统科学与综合研究,2007,23(3):265-267,271.
[14]黄安,杨联安,杜挺,等.基于主成分分析的土壤养分综合评价[J].干旱区研究,2014,(5):819-825.
[15]李旭东.贵州喀斯特高原人口分布的自然环境因素Ⅱ:多元回归分析与地带性研究[J].干旱区研究,2007,24(2):66-72.
[16]涂向阳.模糊数学在海水入侵地下水水质评价中的应用[J].水利学报,2003,34(8):64-69.
[17]陈峰,蒋新,唐访良,等.层次分析法与地理信息系统在农田土壤重金属污染评价中的应用[J].环境污染与防治,2012,34(7):6-8,14.
[18]侯晓丽,冯跃华,吴光辉,等.基于人工神经网络土壤墒情动态预测模型应用研究[J].节水灌溉,2016(7):70-72,76.
[19]李军梅,胡以华,陶小红.基于主成分分析与BP神经网络的识别方法研究[J].红外与激光工程,2005(6):719-723.
[20]张轶莹.基于主成分分析法的BP神经网络基准地价评估研究[D].郑州:河南农业大学,2015.
[21]沈小燕,刘浩学,晏远春.基于主成分分析和BP神经网络的综合安全评价模型[J].安全与环境学报,2009,9(1):180-184.
[22]张淑娟,王凤花,张海红,等.基于主成分分析和BP神经网络的土壤养分近红外光谱检测[J].山西农业大学学报(自然科学版),2009,29(6):483-487.
[23]郭李娜,樊贵盛.基于灰色理论—BP神经网络方法的表层土壤容重预测[J].节水灌溉,2018(2):93-97.
[24]王宏宇,马娟娟,孙西欢,等.基于BP神经网络的土壤水热动态预测模型研究[J].节水灌溉,2017(7):11-15,27.
[25]赵玉杰,师荣光,高怀友,等.基于MATLAB6.x的BP人工神经网络的土壤环境质量评价方法研究[J].农业环境科学学报,2006,25(1):186-189.
[26]宛恒,杨树青.内蒙古东部区膜下滴灌条件下农田水土环境研究[C]∥东北四省区节水增粮高效灌溉技术研究与规模化示范学术成果交流会论文汇编.2017:137-144.
[27]王海霞.模糊神经网络在水质评价中的应用[D].重庆:重庆大学,2002.
[28]许坤鹏,武世亮,马孝义,等.基于主成分分析土壤水分扩散率单一参数模型的BP神经网络模型[J].干旱区地理,2015,38(1):76-82.
[2]李娟,叶胜兰,牛岩.不同模式对盐碱地土壤环境改良效应的研究[J].中国农村水利水电,2018(9):193-198.
[3] Arnebrant K.Changes in microfungal community structure after fertilization of Scots pine forest soil with ammonium nitrate or urea[J].Transactions of the Chinese Society of Agricultural Engineering,1990,22(3):309-312.
[4] Singh J S,Raghubanshi A S,Singh R S,et al. Microbial biomass acts as a source of plant nutrients in dry tropical forest and savanna[J]. Nature,1989(338):499-500.
[5]马祥华,焦菊英.黄土丘陵沟壑区退耕地自然恢复植被特征及其与土壤环境的关系[J].中国水土保持科学,2005,3(2):15-22.
[6]叶存旺,翟巧绒,郭梓娟,等.沙棘-侧柏混交林土壤养分、微生物与酶活性的研究[J].西北林学院学报,2007,22(5):1-6.
[7]侯琳,雷瑞德,王得祥.黄龙山林区封育油松林土壤养分研究[J].西北农林科技大学学报(自然科学版),2007,35(2):63-68.
[8]韩万海,王增丽.不同灌溉模式对土壤水盐分布及棉花产量的影响[J].中国农村水利水电,2018(11):26-29,34.
[9]柴慧祥.内蒙古通辽市滴灌条件下农田水土环境监测评估[D].呼和浩特:内蒙古农业大学,2016.
[10]吴军斌,尹娟,苏振娟,等.滴灌施肥条件下土壤水分运移规律的研究[J].节水灌溉,2018(11):51-55.
[11]田兆顺,董汉章.华北平原瓦碱的特性和形成[J].土壤学报,1965(1):24-38.
[12] Webster R. Quantitative and numerical methods in soil classification and survey[M]. Clarendon Press,1977:1-7.
[13]赵青,王子龙,姜秋香.属性识别理论模型在土壤肥力综合评价中的应用[J].农业系统科学与综合研究,2007,23(3):265-267,271.
[14]黄安,杨联安,杜挺,等.基于主成分分析的土壤养分综合评价[J].干旱区研究,2014,(5):819-825.
[15]李旭东.贵州喀斯特高原人口分布的自然环境因素Ⅱ:多元回归分析与地带性研究[J].干旱区研究,2007,24(2):66-72.
[16]涂向阳.模糊数学在海水入侵地下水水质评价中的应用[J].水利学报,2003,34(8):64-69.
[17]陈峰,蒋新,唐访良,等.层次分析法与地理信息系统在农田土壤重金属污染评价中的应用[J].环境污染与防治,2012,34(7):6-8,14.
[18]侯晓丽,冯跃华,吴光辉,等.基于人工神经网络土壤墒情动态预测模型应用研究[J].节水灌溉,2016(7):70-72,76.
[19]李军梅,胡以华,陶小红.基于主成分分析与BP神经网络的识别方法研究[J].红外与激光工程,2005(6):719-723.
[20]张轶莹.基于主成分分析法的BP神经网络基准地价评估研究[D].郑州:河南农业大学,2015.
[21]沈小燕,刘浩学,晏远春.基于主成分分析和BP神经网络的综合安全评价模型[J].安全与环境学报,2009,9(1):180-184.
[22]张淑娟,王凤花,张海红,等.基于主成分分析和BP神经网络的土壤养分近红外光谱检测[J].山西农业大学学报(自然科学版),2009,29(6):483-487.
[23]郭李娜,樊贵盛.基于灰色理论—BP神经网络方法的表层土壤容重预测[J].节水灌溉,2018(2):93-97.
[24]王宏宇,马娟娟,孙西欢,等.基于BP神经网络的土壤水热动态预测模型研究[J].节水灌溉,2017(7):11-15,27.
[25]赵玉杰,师荣光,高怀友,等.基于MATLAB6.x的BP人工神经网络的土壤环境质量评价方法研究[J].农业环境科学学报,2006,25(1):186-189.
[26]宛恒,杨树青.内蒙古东部区膜下滴灌条件下农田水土环境研究[C]∥东北四省区节水增粮高效灌溉技术研究与规模化示范学术成果交流会论文汇编.2017:137-144.
[27]王海霞.模糊神经网络在水质评价中的应用[D].重庆:重庆大学,2002.
[28]许坤鹏,武世亮,马孝义,等.基于主成分分析土壤水分扩散率单一参数模型的BP神经网络模型[J].干旱区地理,2015,38(1):76-82.