人工神经网络模型在地学研究中的应用进展
|
摘要
近年来,随着人工神经网络(ANNs)自身技术的不断完善,应用ANNs模型成功解决各类地学问题的案例大量出现。通过对其发展历程进行分析发现,20世纪80年代末国际地学分析中已开始融入ANNs技术,国内则滞后1～2年。在地学分析中使用的各类人工神经网络类型中,BP模型应用最广,占到85%以上。在10余年的应用过程中,虽然地学的各个分支学科都移植了一种或数种ANNs模型作为其分析工具,但水文、地质、大气、遥感等领域应用较为广泛。传统地学定量分析中的单变量或多变量预测成为人工神经网络地学模型的主要应用客体。同时,诸如模式识别和过程模拟等也是ANNs模型求解的对象。目前,随着建模经验和知识的积累,地学ANNs模型的发展呈现出多种技术综合集成的态势,遗传算法、小波转换、模拟退火算法以及模糊逻辑等方法与ANNs模型融合,成为解决地学分析中非线性问题的利器。
Neural networks are increasingly popular in geosciences due to big progress in neural network modelling techniques and imperative demands in geosciences. Without assuming parametric relationship, artificial neural networks have the ability to learn patterns of relationships in data from being shown a given set of inputs including combinations of descriptive and quantitative data, generalize or abstract results from imperfect data, and be insensitive to minor variations in input such as noise in the data, missing data, or a few incorrect values. When the neural network is trained appropriately, it generalizes the relationship so that it can be applied to other new data sets. The theoretical basis of this technique is the universal approximation theorem, which states that a multilayer feedforward neural network, such as the radialbasis function or perceptron neural network, is capable of performing any nonlinear inputoutput mapping. In this paper, the applications of ANNs in various branches of geosciences have been examined here. It is found that ANNs are robust tools and alternative approaches for modeling many of the nonlinear processes in geosciences. After appropriate training, they are able to generate satisfactory results for solving many problems such as prediction, classification, pattern recognition and optimization. By counting the 349 ANNsbased papers in geosciences during the 1997-2000, three predominant subjects are hydrology, geology and atmospheric science, and prediction is major modelling purpose. After reviewing the state of the art of geoscience's ANN modelling, author thinks that integrated ANNs modelling framework must be developed in order to deal with more complex nonlinear processes. Such integrated framework consists of nonparametric techniques such as neural network, fuzzy logic, genetic algorithm, simulated annealing algorithm, fractal theory, Cellular automata and wavelet transform etc. It is helpful for selecting appropriate input and output neurons and designing more efficient networks to profoundly understand the linear or nonlinear processes being modeled in geosciences. Important aspects such as physical interpretation of ANN architecture, optimal training data set, adaptive learning, and generalization must be explored further. The merits and limitations of ANN applications have been discussed, and potential research avenues have been explored briefly.
Neural networks are increasingly popular in geosciences due to big progress in neural network modelling techniques and imperative demands in geosciences. Without assuming parametric relationship, artificial neural networks have the ability to learn patterns of relationships in data from being shown a given set of inputs including combinations of descriptive and quantitative data, generalize or abstract results from imperfect data, and be insensitive to minor variations in input such as noise in the data, missing data, or a few incorrect values. When the neural network is trained appropriately, it generalizes the relationship so that it can be applied to other new data sets. The theoretical basis of this technique is the universal approximation theorem, which states that a multilayer feedforward neural network, such as the radialbasis function or perceptron neural network, is capable of performing any nonlinear inputoutput mapping. In this paper, the applications of ANNs in various branches of geosciences have been examined here. It is found that ANNs are robust tools and alternative approaches for modeling many of the nonlinear processes in geosciences. After appropriate training, they are able to generate satisfactory results for solving many problems such as prediction, classification, pattern recognition and optimization. By counting the 349 ANNsbased papers in geosciences during the 1997-2000, three predominant subjects are hydrology, geology and atmospheric science, and prediction is major modelling purpose. After reviewing the state of the art of geoscience's ANN modelling, author thinks that integrated ANNs modelling framework must be developed in order to deal with more complex nonlinear processes. Such integrated framework consists of nonparametric techniques such as neural network, fuzzy logic, genetic algorithm, simulated annealing algorithm, fractal theory, Cellular automata and wavelet transform etc. It is helpful for selecting appropriate input and output neurons and designing more efficient networks to profoundly understand the linear or nonlinear processes being modeled in geosciences. Important aspects such as physical interpretation of ANN architecture, optimal training data set, adaptive learning, and generalization must be explored further. The merits and limitations of ANN applications have been discussed, and potential research avenues have been explored briefly.
引文
[1] ShiChenggang,LiuXila.Applicationofneuralnetworktoearthquakeengineering[J].EarthquakeEngineeringVibration,1991,15(2):39 47.[石成钢,刘西拉.人工神经网络在震害预测中的应用[J].地震工程与工程振动,1991,15(2):39 47.]
[2] MccannDW.Aneuralnetworkshort termloadforecastofsignificantthunderstorms[J].WeatherForecasting,1992,7:525 535.
[3] WuXiao.Applicationofartificialneuralnetworktorainstormforecast[A].In:IntelligentControlandIntelligentAutomation[C].Beijing:SciencePress,1993.[吴晓.人工神经网络在暴雨预测中的应用[A].智能控制与智能自动化[C].北京:科学出版社,1993.]
[4] MacherGL,FullerJD.Back propagationinhydrologicaltimeseriesofforecasting[A].In:SymposiumofInternationalConferenceonStochasticandStatisticalMethodinHydrologyandEnvironmentalEngineering[C].Ontario,Canada:1993.229 242.
[5] GovindarajuRS.ArtificialneuralnetworksinhydrologyI:Preliminaryconcepts[J].JournalofHydrologicEngineering,2000,5(2):115 123.
[6] GovindarajuRS.ArtificialneuralnetworksinhydrologyII:Hydrologicapplications[J].JournalofHydrologicEngineering,372000,5(2):124 137.
[7] GardnerMW,DorlingSR.Artificialneuralnetworks(multilayerperceptron)—Areviewofapplicationsintheatmosphericsciences[J].AtmosphericEnvironment,1998,32(14～15):26272636.
[8] DawsonCW,WilbyRL.Acomparisonofartificialneuralnetworksusedforriverflowforecasting[J].HydrologyandEarthSystemSciences,1999,3(4):529 540.
[9] SnellSS,GopalS,KaufmanRK.Spatialinterpolationofsurfaceairtemperaturesusingartificialneuralnetworks:EvaluatingtheirusefordownscalingGCMs[J].JournalofClimate,2000,13(5):886 895.
[10] BrownWM,GedeonTD,GrovesDI,etal.Anewmethodfromineralprospectivitymapping[J].AustralianJournalofEarthSciences,2000,47(4):757 770.
[11] MinasnyB,McBratneyAB.EvaluationanddevelopmentofhydraulicconductivitypedotransferfunctionforAustraliansoil[J].AustralianJournalofSoilResearch,2000,38(4):905 926.
[12] SpellmanG.AnapplicationofartificialneuralnetworkstothepredictionofsurfaceozoneconcentrationsintheUnitedKingdom[J].Applied Geography,1999,19(2):123 136.
[13] ZhangB,GovindarajuRS.PredictionofwatershedrunoffusingBayesianconceptsandmodularneuralnetworks[J].WaterResourcesResearch,2000,36(3):753 762.
[14] ImrieCE,DurucanS,KorreA.Riverflowpredictionusingartificialnetworks:Generalizationbeyondthecalibrationrange[J].JournalofHydrology,2000,233(1～4):138 153.
[15] MaierHR,DandyGC.Empiricalcomparisonofvariousmethodsfortrainingfeed forwardneuralnetworksforsalinityforecasting[J].WaterResourcesResearch,1999,35(8):25912596.
[16] AnmalaJ,ZhangB,GovindarajuRS.ComparisonofANNsandempiricalapproachesforpredictingwatershedrunoff[J].JournalofWaterResourcesPlanningandManagement,2000,126(3):156 166.
[17] AbuelgasimAA,RossWD,GopalS,etal.Changedetectionusingadaptivefuzzyneuralnetworks:EnvironmentaldamageassessmentaftertheGulfWar[J].RemoteSensingofEnvironment,1999,70(2):208 223.
[18] ZhuChengshan,PanYingjie.AfuzzyneuralnetworkmethodforpredictingtherunoffintoOuyanghaireservoir[J].InternationalJournalofHydroelectricEnergy,2000,18(3):16 18.[朱承山,潘英杰.欧阳海水库径流量预报的模糊神经网络方法[J].水电能源科学,2000,18(3):16 18.]
[19] CaoJie,XieYingqi.Onekindofassociativememorybyneuralnetwork[J].JournalofTropicalMeteorology,1997,13(1):8287.[曹杰,谢应齐.一种基于混沌理论的联想记忆神经网络模型[J].热带气象学报,1997,13(1):82 87.]
[20] NunnariG,NuciforaAFM,RandieriC.Theapplicationofneuraltechniquestothemodellingoftime seriesofatmosphericpollutiondata[J].EcologicalModelling,1998,111(2～3):187 205.
[21] JinLong,LuoYing,MiaoQilong,etal.Forecastmodeloffarmlandsoilmoisturebyartificialneuralnetworks[J].ActaFedologicaSinica,1998,35(1):25 32.[金龙,罗莹,缪启龙,等.农田土壤湿度的人工神经网络预报模型研究[J].土壤学报,1998,35(1):25 32.]
[22] HongWei,WuChengzhen.StudyonsoillosspredictionbyartificialnetworkinsoutheastFujian[J].JournalofSoilErosionandSoilandWaterConservation,1997,3(3):52 57.[洪伟,吴承祯.闽东南土壤流失人工神经网络预报研究[J].土壤侵蚀与水土保持学报,1997,3(3):52 57.]
[23] LiXianbin,DingJing.Combiningneuralnetworkforecastsonwavelet transformedseries[J].JournalofHydraulicEngineering,1999,2:1 4.[李贤彬,丁晶.基于子波变换序
[2] MccannDW.Aneuralnetworkshort termloadforecastofsignificantthunderstorms[J].WeatherForecasting,1992,7:525 535.
[3] WuXiao.Applicationofartificialneuralnetworktorainstormforecast[A].In:IntelligentControlandIntelligentAutomation[C].Beijing:SciencePress,1993.[吴晓.人工神经网络在暴雨预测中的应用[A].智能控制与智能自动化[C].北京:科学出版社,1993.]
[4] MacherGL,FullerJD.Back propagationinhydrologicaltimeseriesofforecasting[A].In:SymposiumofInternationalConferenceonStochasticandStatisticalMethodinHydrologyandEnvironmentalEngineering[C].Ontario,Canada:1993.229 242.
[5] GovindarajuRS.ArtificialneuralnetworksinhydrologyI:Preliminaryconcepts[J].JournalofHydrologicEngineering,2000,5(2):115 123.
[6] GovindarajuRS.ArtificialneuralnetworksinhydrologyII:Hydrologicapplications[J].JournalofHydrologicEngineering,372000,5(2):124 137.
[7] GardnerMW,DorlingSR.Artificialneuralnetworks(multilayerperceptron)—Areviewofapplicationsintheatmosphericsciences[J].AtmosphericEnvironment,1998,32(14～15):26272636.
[8] DawsonCW,WilbyRL.Acomparisonofartificialneuralnetworksusedforriverflowforecasting[J].HydrologyandEarthSystemSciences,1999,3(4):529 540.
[9] SnellSS,GopalS,KaufmanRK.Spatialinterpolationofsurfaceairtemperaturesusingartificialneuralnetworks:EvaluatingtheirusefordownscalingGCMs[J].JournalofClimate,2000,13(5):886 895.
[10] BrownWM,GedeonTD,GrovesDI,etal.Anewmethodfromineralprospectivitymapping[J].AustralianJournalofEarthSciences,2000,47(4):757 770.
[11] MinasnyB,McBratneyAB.EvaluationanddevelopmentofhydraulicconductivitypedotransferfunctionforAustraliansoil[J].AustralianJournalofSoilResearch,2000,38(4):905 926.
[12] SpellmanG.AnapplicationofartificialneuralnetworkstothepredictionofsurfaceozoneconcentrationsintheUnitedKingdom[J].Applied Geography,1999,19(2):123 136.
[13] ZhangB,GovindarajuRS.PredictionofwatershedrunoffusingBayesianconceptsandmodularneuralnetworks[J].WaterResourcesResearch,2000,36(3):753 762.
[14] ImrieCE,DurucanS,KorreA.Riverflowpredictionusingartificialnetworks:Generalizationbeyondthecalibrationrange[J].JournalofHydrology,2000,233(1～4):138 153.
[15] MaierHR,DandyGC.Empiricalcomparisonofvariousmethodsfortrainingfeed forwardneuralnetworksforsalinityforecasting[J].WaterResourcesResearch,1999,35(8):25912596.
[16] AnmalaJ,ZhangB,GovindarajuRS.ComparisonofANNsandempiricalapproachesforpredictingwatershedrunoff[J].JournalofWaterResourcesPlanningandManagement,2000,126(3):156 166.
[17] AbuelgasimAA,RossWD,GopalS,etal.Changedetectionusingadaptivefuzzyneuralnetworks:EnvironmentaldamageassessmentaftertheGulfWar[J].RemoteSensingofEnvironment,1999,70(2):208 223.
[18] ZhuChengshan,PanYingjie.AfuzzyneuralnetworkmethodforpredictingtherunoffintoOuyanghaireservoir[J].InternationalJournalofHydroelectricEnergy,2000,18(3):16 18.[朱承山,潘英杰.欧阳海水库径流量预报的模糊神经网络方法[J].水电能源科学,2000,18(3):16 18.]
[19] CaoJie,XieYingqi.Onekindofassociativememorybyneuralnetwork[J].JournalofTropicalMeteorology,1997,13(1):8287.[曹杰,谢应齐.一种基于混沌理论的联想记忆神经网络模型[J].热带气象学报,1997,13(1):82 87.]
[20] NunnariG,NuciforaAFM,RandieriC.Theapplicationofneuraltechniquestothemodellingoftime seriesofatmosphericpollutiondata[J].EcologicalModelling,1998,111(2～3):187 205.
[21] JinLong,LuoYing,MiaoQilong,etal.Forecastmodeloffarmlandsoilmoisturebyartificialneuralnetworks[J].ActaFedologicaSinica,1998,35(1):25 32.[金龙,罗莹,缪启龙,等.农田土壤湿度的人工神经网络预报模型研究[J].土壤学报,1998,35(1):25 32.]
[22] HongWei,WuChengzhen.StudyonsoillosspredictionbyartificialnetworkinsoutheastFujian[J].JournalofSoilErosionandSoilandWaterConservation,1997,3(3):52 57.[洪伟,吴承祯.闽东南土壤流失人工神经网络预报研究[J].土壤侵蚀与水土保持学报,1997,3(3):52 57.]
[23] LiXianbin,DingJing.Combiningneuralnetworkforecastsonwavelet transformedseries[J].JournalofHydraulicEngineering,1999,2:1 4.[李贤彬,丁晶.基于子波变换序
版权所有：© 2023 中国地质图书馆 中国地质调查局地学文献中心