精准农业生产中若干智能决策问题研究
|
摘要
围绕国家863项目“农业知识网格研究与应用”(2006AA10Z245)和“玉米精准作业系统研究与应用”(2006AA10A309),针对精准农业生产中精准施肥模型构建、地力分级、产量预测、管理区划分等方面存在的难题,基于神经网络、贝叶斯网、模糊聚类等,提出了一组新的智能决策方法;集成地理信息系统和专家系统,构建了精准农业智能决策系统。具体研究内容如下:
(1)提出一种基于AP聚类和Lagrange乘子的神经网络集成方法,并考察了子网络个数和聚类数目对神经网络拟合精度的影响。然后针对两种不同的玉米田进行实验,采用上述的神经网络集成方法,建立了精准施肥模型。实验结果表明,本文提出的神经网络集成方法优于已有的神经网络集成方法。同时,基于神经网络集成的施肥模型不仅优于传统施肥模型,而且优于基于单个神经网络的施肥模型,能有效地指导精准施肥。
(2)建立一组基于贝叶斯网络的精准农业智能决策模型。首先,提出一种简化的贝叶斯网络模型,构建了精准农业生产技术决策网;然后,基于贝叶斯网络学习方法,生成了产量预测模型;最后,基于贝叶斯网络分类器,生成了地力分级模型。实验结果表明,贝叶斯网能有效地处理精准农业生产中的不确定信息,能够较好地模拟多种因素之间存在的复杂因果关系。
(3)提出一组管理区划分方法。针对不同土壤养分对地力贡献率的不同,提出一种基于加权模糊聚类的管理区划分方法。基于不同肥料效应下的产量,提出一种基于灰色关联度的管理区划分方法;针对土壤养分在不同方向具有不同变异度的情况,提出一种基于有向变异度的管理区划分方法;实验结果表明,上述方法划分的管理区比较合理,基本符合农业生产实际。
(4)以玉米生产的精准化、智能化和网络化为目标,集成地理信息系统(GIS)和专家系统(ES),研制出基于网络的面向两类用户的玉米精准作业智能决策系统。一方面,该系统面向以GPS网格为操作单元采用变量施肥机的用户;另一方面,该系统面向以农户地块为操作单元采用普通施肥机的用户。
总之,以上方法能较好地处理精准农业生产中的时空信息和不确定性信息,其效果明显优于传统方法,对精准农业智能决策研究具有一定的理论意义和应用价值。
Precision agriculture is the effective shortcut to implement Agricultural Sustainable development. Getting accurate and reasonable decision by dealing with agricultural data and knowledge is the core content of precision agriculture. With the development of agriculture and the progress of technology, some new cases and problems appeared. Traditional methods of treating data can’t meet the need of precision, digitization and intellectualization of agricultural production. In this thesis, guided by agricultural systematology and taken maize precision production as research object, according to the problem existing in construction of fertilization model,evaluation of land capability, yield forecasting, division of management zones and developing of agricultural intelligent decision system, based on Bayesian network, fuzzy clustering, neural network ensemble, expert system, etc, some new methods for precision agriculture decision are proposed, which result in good effect.
1 . According to the complexity and uncertainty of precision agriculture information the agricultural system theory is introduced. Then the intelligent decision technology architecture is constructed.
(1)Guided by agricultural system theory, the requirement of precision agriculture decision is analyzed. The solving of precision agriculture problem is divided into three layers, namely data, knowledge and decision.
(2)It is pointed that spatio-temperal quality and uncertainty is the potential nature of precision agriculture information. Then the precision agriculture technology architecture with the shape of hourglass is constructed.
(3)The method of combining intelligent decision technology with precision agriculture is given. Finally,Bayesian network,neural network, etc used in this paper are introduced.
2.A novel neural network ensemble method is proposed and used to construct precision fertilization model.
(1)A neural network ensemble method based on AP clustering and Lagrange Multiplier is proposed. In which, the method of sampling with replacement is used to produce networks and a novel formula is defined to measure the similarity of networks. By the AP clustering algorithm,networks with high precision and greater diversity are selected. Then by the Lagrange Multiplier(LM) method, these networks are combined. The experiment on the standard dataset shows that, the algorithm has higher accuracy and stronger generalization than the single neural network. Furthermore, as a linear weighted ensemble method, LM ensemble is better than Forecasting Effective Measure(FEM)ensemble and equal weight.
(2) Base on the above research,two kinds of fertilization model based on neural network are proposed. One is the 4-x-3 model,the other is 6-x-1 model. The result shows that both models can simulate the nonlinear relation existing in soil nutrient, fertilization rate and yield. Especially,the 6-x-1 model not only can compute fertilization rate precisely but also can forecast yield. The practice shows that neural network ensemble based fertilization model is better than traditional fertilization models and existing neural network based fertilization models.
3.Bayesian network is a strong tool of dealing with uncertainty knowledge, which has the advantage of bi-direction reasoning, fusion of prior knowledge, etc. In this thesis, Bayesian network is used in precision agriculture decision.
(1)A method of manual constructing Bayesian network through integrating expert experience and“noisy sum”model is proposed. Then the method is used to construct decision network for precision agriculture, which can make decisions by different utility.
(2)A Bayesian network for yield forecasting is constructed. Independence analysis shows that when the N,P ,K is known,the yield is determined without knowing H2O and organic matter. Sensitivity analysis shows that plant height in crop trait and P in soil are very important factors impacting yield.
(3)Land classification model based on Bayesian classifier is constructed. The results show that both TAN model and na?ve Bayesian model have higher accuracy rate(larger than 82%),so they can be adopted in actual production. On the other hand, the accuracy rate of TAN model is a little larger than na?ve Bayesian model, which verifies that there exists correlation to some extent among the attributes affecting land capability. Finally, by sensitivity analysis for the na?ve Bayesian model, it is concluded that salinity, rainfall and terrain are important factors affecting land capability.
4.Three methods of dividing management zones are proposed.
(1)A division method based on directed variation degree is proposed. The results show that the directed variation degree method is accordance with the practice that the different variation degree exists in the different direction in the field.
(2)A division method based on grey association degree is proposed. In this method, yield data under different treatments in multi points is analyzed by grey association degree, then the management zones is divided by the interval of grey association degree. The results show that this method is suitable for the fertilizer effect data and the situation of scare data.
(3)A division method based on weighted fuzzy clustering is proposed. In this algorithm, the determination of the weight of different soil nutrients is a key step. Two methods are used to determine the weight, one is the primary component analysis(PCA), the other is coefficient of variation (C.V).To verify above methods, at the experimental site, a field in Yushu city in Jilin province, N, P, K were sampled, then the weighted fuzzy clustering algorithm was used and the number of the management zones is determined. Then the effectiveness of the weighted fuzzy clustering method is validated by comparing with the traditional fuzzy clustering method. The results shows more practical and can be used to direct precision fertilization.
5.Aiming to the precision, intelligence and networking of maize production,a intelligent decision system for maize precision production based on web is developed. The features of the system are as follows::
(1)The system integrates GIS and ES,which makes it deal with spatial information and uncertainty information.
(2)The system integrates two kinds of precision working mode. One is based on GPS grid,and the other is based on farmer field.
The practice results show that the application of the system increases the yield, decreases the price and lesson the pollution.
(1)提出一种基于AP聚类和Lagrange乘子的神经网络集成方法,并考察了子网络个数和聚类数目对神经网络拟合精度的影响。然后针对两种不同的玉米田进行实验,采用上述的神经网络集成方法,建立了精准施肥模型。实验结果表明,本文提出的神经网络集成方法优于已有的神经网络集成方法。同时,基于神经网络集成的施肥模型不仅优于传统施肥模型,而且优于基于单个神经网络的施肥模型,能有效地指导精准施肥。
(2)建立一组基于贝叶斯网络的精准农业智能决策模型。首先,提出一种简化的贝叶斯网络模型,构建了精准农业生产技术决策网;然后,基于贝叶斯网络学习方法,生成了产量预测模型;最后,基于贝叶斯网络分类器,生成了地力分级模型。实验结果表明,贝叶斯网能有效地处理精准农业生产中的不确定信息,能够较好地模拟多种因素之间存在的复杂因果关系。
(3)提出一组管理区划分方法。针对不同土壤养分对地力贡献率的不同,提出一种基于加权模糊聚类的管理区划分方法。基于不同肥料效应下的产量,提出一种基于灰色关联度的管理区划分方法;针对土壤养分在不同方向具有不同变异度的情况,提出一种基于有向变异度的管理区划分方法;实验结果表明,上述方法划分的管理区比较合理,基本符合农业生产实际。
(4)以玉米生产的精准化、智能化和网络化为目标,集成地理信息系统(GIS)和专家系统(ES),研制出基于网络的面向两类用户的玉米精准作业智能决策系统。一方面,该系统面向以GPS网格为操作单元采用变量施肥机的用户;另一方面,该系统面向以农户地块为操作单元采用普通施肥机的用户。
总之,以上方法能较好地处理精准农业生产中的时空信息和不确定性信息,其效果明显优于传统方法,对精准农业智能决策研究具有一定的理论意义和应用价值。
Precision agriculture is the effective shortcut to implement Agricultural Sustainable development. Getting accurate and reasonable decision by dealing with agricultural data and knowledge is the core content of precision agriculture. With the development of agriculture and the progress of technology, some new cases and problems appeared. Traditional methods of treating data can’t meet the need of precision, digitization and intellectualization of agricultural production. In this thesis, guided by agricultural systematology and taken maize precision production as research object, according to the problem existing in construction of fertilization model,evaluation of land capability, yield forecasting, division of management zones and developing of agricultural intelligent decision system, based on Bayesian network, fuzzy clustering, neural network ensemble, expert system, etc, some new methods for precision agriculture decision are proposed, which result in good effect.
1 . According to the complexity and uncertainty of precision agriculture information the agricultural system theory is introduced. Then the intelligent decision technology architecture is constructed.
(1)Guided by agricultural system theory, the requirement of precision agriculture decision is analyzed. The solving of precision agriculture problem is divided into three layers, namely data, knowledge and decision.
(2)It is pointed that spatio-temperal quality and uncertainty is the potential nature of precision agriculture information. Then the precision agriculture technology architecture with the shape of hourglass is constructed.
(3)The method of combining intelligent decision technology with precision agriculture is given. Finally,Bayesian network,neural network, etc used in this paper are introduced.
2.A novel neural network ensemble method is proposed and used to construct precision fertilization model.
(1)A neural network ensemble method based on AP clustering and Lagrange Multiplier is proposed. In which, the method of sampling with replacement is used to produce networks and a novel formula is defined to measure the similarity of networks. By the AP clustering algorithm,networks with high precision and greater diversity are selected. Then by the Lagrange Multiplier(LM) method, these networks are combined. The experiment on the standard dataset shows that, the algorithm has higher accuracy and stronger generalization than the single neural network. Furthermore, as a linear weighted ensemble method, LM ensemble is better than Forecasting Effective Measure(FEM)ensemble and equal weight.
(2) Base on the above research,two kinds of fertilization model based on neural network are proposed. One is the 4-x-3 model,the other is 6-x-1 model. The result shows that both models can simulate the nonlinear relation existing in soil nutrient, fertilization rate and yield. Especially,the 6-x-1 model not only can compute fertilization rate precisely but also can forecast yield. The practice shows that neural network ensemble based fertilization model is better than traditional fertilization models and existing neural network based fertilization models.
3.Bayesian network is a strong tool of dealing with uncertainty knowledge, which has the advantage of bi-direction reasoning, fusion of prior knowledge, etc. In this thesis, Bayesian network is used in precision agriculture decision.
(1)A method of manual constructing Bayesian network through integrating expert experience and“noisy sum”model is proposed. Then the method is used to construct decision network for precision agriculture, which can make decisions by different utility.
(2)A Bayesian network for yield forecasting is constructed. Independence analysis shows that when the N,P ,K is known,the yield is determined without knowing H2O and organic matter. Sensitivity analysis shows that plant height in crop trait and P in soil are very important factors impacting yield.
(3)Land classification model based on Bayesian classifier is constructed. The results show that both TAN model and na?ve Bayesian model have higher accuracy rate(larger than 82%),so they can be adopted in actual production. On the other hand, the accuracy rate of TAN model is a little larger than na?ve Bayesian model, which verifies that there exists correlation to some extent among the attributes affecting land capability. Finally, by sensitivity analysis for the na?ve Bayesian model, it is concluded that salinity, rainfall and terrain are important factors affecting land capability.
4.Three methods of dividing management zones are proposed.
(1)A division method based on directed variation degree is proposed. The results show that the directed variation degree method is accordance with the practice that the different variation degree exists in the different direction in the field.
(2)A division method based on grey association degree is proposed. In this method, yield data under different treatments in multi points is analyzed by grey association degree, then the management zones is divided by the interval of grey association degree. The results show that this method is suitable for the fertilizer effect data and the situation of scare data.
(3)A division method based on weighted fuzzy clustering is proposed. In this algorithm, the determination of the weight of different soil nutrients is a key step. Two methods are used to determine the weight, one is the primary component analysis(PCA), the other is coefficient of variation (C.V).To verify above methods, at the experimental site, a field in Yushu city in Jilin province, N, P, K were sampled, then the weighted fuzzy clustering algorithm was used and the number of the management zones is determined. Then the effectiveness of the weighted fuzzy clustering method is validated by comparing with the traditional fuzzy clustering method. The results shows more practical and can be used to direct precision fertilization.
5.Aiming to the precision, intelligence and networking of maize production,a intelligent decision system for maize precision production based on web is developed. The features of the system are as follows::
(1)The system integrates GIS and ES,which makes it deal with spatial information and uncertainty information.
(2)The system integrates two kinds of precision working mode. One is based on GPS grid,and the other is based on farmer field.
The practice results show that the application of the system increases the yield, decreases the price and lesson the pollution.
引文
[1] Richard E. Plant.Site-specific management: the application of information technology to crop production. Computers and Electronics in griculture[J] .2001(30): 9–29.
[2] Yanbo Huang, Yubin Lan, Steven J. Thomson. Development of soft computing and applications in agricultural and biological engineering[J]. Computers and Electronics in Agriculture. 2010 (71) :107–127.
[3]付强.数据处理方法及农业应用[M].北京:科学出版社.2006.
[4]张颖,刘艳秋.软计算方法[M].北京:科学出版社,2002.
[5]何勇,张淑娟,方慧.基于人工神经网络的田间信息插值方法研究[J].农业工程学报,2004,20(3):120-123.
[6]付强,王志良,梁川.自组织竞争人工神经网络在土壤分类中的应用[J].水土保持通报,2002,22(1):39-43.
[7] Dragoljub Pokrajac,Zoran Obradovic.A neural network-based method for site-specific fertilization recommendation[A]. California USA:ASAE Annual International Meeting, Sacramento,2001.
[8]腾青芳,秦春林,党建武,等.利用神经网络建立土壤施肥模型的应用研究[J].兰州铁道学院学报(自然科学版)[J].2002,21(4):54-57.
[9]马成林,吴才聪,张书慧.基于数据包络分析和人工神经网络的变量施肥决策方法研究[J].农业工程学报,2004,20(2): 152-155.
[10]兰维娟,毛鹏军,杜东亮,等.基于径向基函数网络的变量施肥决策研究[J].安徽农业科学, 2007,35(21):6505-6507.
[11]冯沈萍等.运用人工神经网络预测一代粘虫发生程度的研究[J].微型电脑应用,1999(15):54-56
[12]马飞,许晓风,张夕林,等.神经网络预警系统及其在害虫预测中的应用.昆虫知识,2002,39(2):115-119..
[13] Drummond,S.Joshi,A.Sudduth,K.A.Application of neural networks: precision farming[A].Neural Networks Proceedings.1998,211-215.
[14] P.Larranaga,S.Moral.Probabilistic graphical models in artificial intelligence.Applied soft computing[J].2008, doi:10.1016/j.asoc.2008.01.003.
[15]张连文,郭海鹏.贝叶斯网引论[M].北京:科学出版社,2006.
[16] F.trai,A bayesian network for predicting yield response of winter wheat to fungicide programs[J].Computers and electronics in agriculture,1996:111-121.
[17] Kristian Kristensen,Ilse A.Rasumussen.The use of a Bayesian network in the design of a decision support system for growing malting barley without use of pestcides[J].Computers and electronics in agriculture.2002:197-217.
[18] Yonghua Qu,Jindi Wang,Huawei Wan,et al.A Bayesian network algorithm for retrieving characterization of land surface vegetation.Remote Sensing of Environmen[J].2008(112),613-622.
[19]于合龙,陈桂芬,董旭初.基于贝叶斯网的专家系统平台的开发与应用[J].黑龙江大学自然科学学报,2008,25(4):492-497.
[20]邓聚龙.农业系统灰色理论与方法[M].济南:山东科学技术出版社.1988.
[21]郭瑞林.作物灰色育种学[M].北京:中国农业科技出版社.1995.
[22]王占中,郭瑞林,齐光荣,等.基于Java的作物灰色育种智能决策系统研究.安徽农业科学, 2008,36(13):5669-5670,5675.
[23]陈桂芬,曹丽英,赵月玲.Fuzzy聚类分析在土壤养分分类中的应用[J].玉米科学,2008,16(2):118-121.
[24] Li yan,Shui Zhou,Li Feng, Li Hong-Yi.Delineation of site-specific management zones using fuzzy clustering analysis in acoastal saline land[J].Computers and electronics in agriculture,2007(56):174-186.
[25]陈彦,吕新.基于FCM的绿洲农田养分管理区研究[J].中国农业科学.2008,41(7):2016-2024.
[26]张璐,刘淑英,王平.基于模糊数学的农用地质量综合评价模型[J].广东农业科学,2010,2:207-2-9,215.
[27]王春平,张改生,张伟,等.应用模糊综合评判和灰色关联度分析评估小麦新品种[J].西北农林科技大学学报(自然科学版),2005,33(8):126-130)
[28]唐惠丽,周炼清,叶基瑶.基于模糊推理的农业病虫害专家系统[J].浙江农业学报,2009 ,21(5):506～509.
[29] Ambuel,J.R.,Colvin,et al. A fuzzy logic yield simulator for prescription farming[J].Transactions of ASAE.1994,37(6):1999-2009.
[30] Yang,C.C.,Prasher,et al.Development of an imag processing system and a fuzzy algorithm for site-specific herbicide applications.precision agriculture[J].Precision Agriculture,2003,4(1):5-18.
[31]唐启义,冯明光.DPS数据处理系统-实验设计、统计分析及数据挖掘[M].科学出版社,2007.1.
[32]全国农业技术推广服务中心.耕地地力指南评价指南[M].中国农业科学技术出版社,2006.
[33]赵春江.农业智能系统[M].北京:科学出版社.2009.1
[34]曹卫星.数字农作技术[M].北京:科学出版社.2008.8
[35] Y.cohen,A.Cohen,A.Hetzron,et al.Spatial decision support system for Medfly control in citrus[J]. computers and electronics in agriculture.2008,(62):107-117.
[36]陈蓉蓉.精确农作决策支持系统的研究[D].南京农业大学,2003.
[37]陈云坪,赵春江,王秀,等.基于知识模型与WebGIS的精准农业处方智能生成系统研究[J].中国农业科学,2007,40(6):1190-1197.
[38] Edson Murakami,Antomia M.Saraiva,Luiz C.M.Ribeiro Junior,etc.An infrastructure for the development of distributed service-oriented information systems for precision agriculture[J].computers and electronics in agriculture.2007(58):37-48.
[39] Allan Leck Jensen.Building a web-based information system for variety selection in field crops:objectives and results[J].Computers and Electronics in Agriculture.2001(32):195–211.
[40]高亮之.农业系统学基础[M].南京:江苏科学技术出版社,1993.
[41]罗锡文,张泰岭,洪添胜.“精细农业”技术体系及其应用[J].2001,32(2):103-106.
[42]赵春江,薛绪掌,王秀.精准农业技术体系的研究进展与展望[J].农业工程学报,2003,19(4):7-12.
[43]曹宏鑫,张春雷,金之庆,等.数字化栽培的框架与技术体系探讨[J].耕作与栽培,2005,(3):4-7,40.
[44]邬伦,刘瑜,等.地理信息系统-原理、方法与应用[M].北京:科学出版社,2001.
[45] N.A. Lorentzos , A.B. Sideridis, C.P. Yialouris ,et al.An integrated spatiotemporal system[J]. Computers and Electronics in Agriculture. 1999 (22) :233–242.
[46]王生生,刘大有,欧阳继红,等.数字农业时空信息管理平台[J].计算机应用研究,2007,24(5):163-165,185.
[47]张伟,数字农业空间信息管理平台开发研究[D].华东师范大学.2004.
[48]王娟,查良松.GIS与空间数据挖掘集成在农业中的应用[J].安徽农业科学,2008,36(29):12974-12975,12985.
[49]陈桂芬.面向精准农业的空间数据挖掘技术研究与应用[D].吉林大学.2009.
[50] MATTHEW L. ADAMS,U SIMON COOK, ROBERT CORNER.Managing Uncertainty in Site-Specific Management: What is the Best Model[J]. Precision Agriculture,2000,( 2): 39-54.
[51]刘思峰,党耀国,方志耕.灰色系统理论及其应用(第三版)[M].北京:科学出版社,2008.
[52] Dimitrios Moshou , Els Vrindts, Bart De Ketelaere. A neural network based plant classifier[J]. Computers and Electronics in Agriculture. 2001,(31): 5–16.
[53] J.A. Marchant, C.M. Onyango. Comparison of a Bayesian classifier with a multilayer feed-forward neural network using the example of plant/weed/soil discrimination[J]. Computers and Electronics in Agriculture. 2003 (39):3-22.
[54] Hansen L K,Salamon P.Neural network ensembles[C].IEEE Trans Pattern Analysis and Machine Intelligence,1990,12(10):993- 1001.
[55] Hansen L K, Salamon P. Neural network ensembles[C]. IEEE Trans PatternAnalysis and Machine Intelligence, 1990, 12 (10): 993- 1001.
[56] Sollich P, Krogh A. Learning with ensembles: How over-fitting can be useful[J]. In: TouretzkyD, MozerM , HasselmoMeds. Advances in Neural Information Processing Systems 8, Cambridge, MA: MIT Press, 1996. 190- 196.
[57] Schapire R E. The strength of weak learn ability[J]. Machine Learning, 1990, 5 (2): 197- 227.
[58] Breiman L. Bagging predictors[J]. Machine Learning, 1996, 24 (2) : 123- 140
[59] G. Peter Zhang: A neural network ensemble method with jittered training data for time series forecasting[J]. Inf. Sci. 177(23): 5329-5346 (2007).
[60] Zhi-Hua Zhou, Jianxin Wu, Wei Tang. Ensembling neural networks: Many could be better than all[J]. Artif. Intell.137(1-2): 239-263 (2002).
[61]吴建鑫,周志华,沈学华,等.一种选择性神经网络集成构造方法[J].计算机研究与发展.2000,37(9):1039-1044
[62] Zhi-Hua Zhou, Jianxin Wu, Yuan Jiang, Shifu Chen.Genetic Algorithm based Selective Neural Network Ensemble[C]. IJCAI 2001: 797-802.
[63] FU Qiang, HU Shang-xu, ZHAO Sheng-ying. Clustering-based selective neural network ensemble[J]. Journal of Zhejiang University Science,2005, 6A(5): 387-392.
[64]李凯,黄厚宽.一种基于聚类技术的选择性神经网络集成方法[J].计算机研究与发展.2005,42(4):594-598.
[65] Haixia Chen, Senmiao Yuan, Kai Jiang. Selective Neural Network Ensemble Based on Clustering[M]. ISNN (1) 2006: 545-550.
[66] Jian Lin, Bangzhu Zhu. A Novel Kernel Clustering Algorithm Based Selective Neural Network Ensemble Model for Economic Forecasting[C]. ISICA 2007:310-315.
[67] Jian Lin, Bangzhu Zhu. A novel neural network ensemble system for economic forecasting[C].ICIC 2007,CCIS 2, 2007: 1227-1233.
[68] Lean Yu, Shouyang Wang, Kin Keung Lai. A novel nonlinear ensemble forecasting model incorporating GLAR and ANN for foreign exchange rates[J]. Computers & OR 32: 2523-2541 (2005).
[69] Kin Keung Lai, Lean Yu, Shouyang Wang,Wei Huang. A Novel Nonlinear Neural Network Ensemble Model for Financial Time Series Forecasting[C]. International Conference on Computational Science (1) 2006: 790-793.
[70] Frey B J, Dueck D. Clustering by Passing Messages between Data Points[J]. Science, 2007, 315(5814): 972-976.
[71]高祥照,胡克林,郭炎,等.土壤养分与作物产量的空间变异特征与精确施肥[J].中国农业科学.2002:35(6):660-666.
[72]张书慧,马成林,等.变量施肥对玉米产量及土壤养分影响的实验[J].农业工程学报.2006,22(8):64-67.
[73]侯彦林,陈守伦.施肥模型研究综述[J].土壤通报.2004,35(4):493-501.
[74]王圣瑞,陈新平,高祥照等.“3414”肥料试验模型拟合的探讨[J].植物营养与肥料学报.2002,8(4) :409～413.
[75]刘光启.农业速查速算手册(上)--农业常用技术资料、土壤和肥料、农业设施卷[M].化学工业出版社.2008.
[76] Stuart Russell,Peter Norvig.Artificial Intelligence—A modern approach[M](影印版).北京:清华大学版社,2006.
[77] Joseph Giarratano,Gary Riley著,印鉴,刘星成,汤庸译.专家系统原理与编程[M].北京:机械工业出版社.2000.
[78]项利国,沈益民,何勇,等.粗糙集理论在粮食产量影响因素分析中的应用[J].浙江大学学报(农业与生命科学版),2003,29(3):343-346.
[79]于晓秋,葛家麒,刘长海.组合预测方法在黑龙江垦区大豆产量预测中的应用[J].数学的实践与认识,2007,37(24):27-32.
[80]王波,郭夜白,高来斌,等.最优加权预测在吉林省粮食产量预测中的应用[J].吉林农业大学学报. 2008,30(5):760-763.
[81]张谷丰等.基于WebGIS的农作物病虫害预警系统[J].农业工程学报,2007(23):176-181.
[82]段项锁,薛正平,杨星卫,等.精准农业土壤养分与水稻产量关系模型研究[J].中国生态农业学报.2003,11(1):62-65.
[83]薛正平,邓华,杨星卫,等.基于决策树和图层叠置的精准农业产量图分析方法[J].农业工程学报.2006,22(8):140-144.
[84] Georg Ru?, Rudolf Kruse1, Martin Schneider,et al. Data Mining with Neural Networks for Wheat Yield Prediction[J]. Advances in Data Mining.Medical Applications,E-Commerce, Marketing,and Theoretical Aspects. Springer Berlin/Heidelberg.2008.47-56.
[85] J. Cheng, R. Greiner, J. Kelly, D. Bell, W. R. Liu. Learning Bayesian networks from data: An information-theory based approach[J]. Artificial Intelligence, 2002(137): 43-90.
[86]陈争光,蔡德利.模糊物元分析法在耕地地力评价中的应用[J].安徽农学通报, 2008,14(11):55-56,233.
[87]田剑,胡月明,王长委,等.聚类支持下决策树模型在耕地评价中的应用[J].农业工程学报,2007,23 (12):58-63.
[88]周颜军,王双成,王辉.基于贝叶斯网络的分类器研究[J].东北师大学报自然科学版.2003,35(2):21-27.
[89]谢高地,陈沈斌.环境的空间连续变异与精准农业[M].北京:气象出版社.2005.7.
[90]宋新山,邓伟,张琳.MATLAB在环境科学中的应用[M].北京:化学工业出版社.2008.2.
[91]李翔,潘瑜春,赵春江,等.基于多年产量数据的精准逐农业管理分区提取与尺度效应评价[J].中国农业科学,2005,38(9):1825-1833.
[92]李艳,史舟,吴次方,等.基于多源数据的盐碱地精确农作管理分区研究[J].农业工程学报,2007,23(8):84-89.
[93]李翔,潘瑜春,马景宇,等.基于多种土壤养分的精准管理分区方法研究[J].土壤学报.2007,44(1):14-20.
[94] R.Khosla,D.Inman,D.G.Westfall,R.M.Reich,M.Frasier,M.Mzuku,B.Koch,A.Hornung.A synthesis of multi-disciplinary research in precision agriculture:site-specific management zones in the semi-arid western Great Plains of the USA[J].Precision agriculture.2008(9):85-100.
[95]郭龙珠,吕纯波,罗玉峰.应用因子分析法遴选水稻灌溉管理分区关键指标[J].农业系统科学与综合研究,2008,24(3):316-318.
[96]王子龙,付强,姜秋香.基于粒子群算法的土壤养分管理分区[J].农业工程学报,2008(10):80-84.
[97] Rodrigo A.Ortega,Oscar A. Santibanez.Determination of management zones in corn based on soil fertility[J]. Computers and electronics inagriculture. 2007(58):49-59.
[98]李艳双,曾珍香,张闽,等.主成分分析法在多指标综合评价方法中的应用[J].河北工业大学学报. 1999,28(1):94-97.
[99]陈贤,吴兴恩,杨德,等.主成分权重法在番茄果实商品性综合评价上的应用探讨[J].吉林农业科学.2008,33(4):49- 52.
[100]李柏年.加权模糊C-均值聚类[J].模糊系统与数学.2007,21(1):106-110.
[101]任军.气田开发方案模糊加权聚类分析[J].天然气工业.1993,14(2):35-37.
[102]谢季坚,刘承平.模糊数学方法及应用(第三版)[M].武汉:华中科技大学出版社.2005.
[103]李鸿吉.模糊数学基础及实用算法[M].北京:科学出版社.2005.
[104]李洁,高新波,焦李成.基于特征加权的模糊聚类新算法[J].电子学报.2006, 34( 1): 89-92.
[105] Sheng-sheng WANG, Da-you LIU , Spatio-temporal Database with Multi-granularities[C].Advances In Web-Age Information Management (WAIM 2004):PROCEEDINGS LECTURE NOTES IN COMPUTER SCIENCE 3129(LNCS):137-146 2004.
[106]邵维忠,杨芙清.面向对象的系统分析(第2版)[M].北京:清华大学出版社.2006.
[107]徐锋,陈暄.UML面向对象建模基础[M].北京:中国水利水电出版社.2006.
[108] Michael Negnevitsky.人工智能-智能系统指南[M].北京:机械工业出版社.2005.
[2] Yanbo Huang, Yubin Lan, Steven J. Thomson. Development of soft computing and applications in agricultural and biological engineering[J]. Computers and Electronics in Agriculture. 2010 (71) :107–127.
[3]付强.数据处理方法及农业应用[M].北京:科学出版社.2006.
[4]张颖,刘艳秋.软计算方法[M].北京:科学出版社,2002.
[5]何勇,张淑娟,方慧.基于人工神经网络的田间信息插值方法研究[J].农业工程学报,2004,20(3):120-123.
[6]付强,王志良,梁川.自组织竞争人工神经网络在土壤分类中的应用[J].水土保持通报,2002,22(1):39-43.
[7] Dragoljub Pokrajac,Zoran Obradovic.A neural network-based method for site-specific fertilization recommendation[A]. California USA:ASAE Annual International Meeting, Sacramento,2001.
[8]腾青芳,秦春林,党建武,等.利用神经网络建立土壤施肥模型的应用研究[J].兰州铁道学院学报(自然科学版)[J].2002,21(4):54-57.
[9]马成林,吴才聪,张书慧.基于数据包络分析和人工神经网络的变量施肥决策方法研究[J].农业工程学报,2004,20(2): 152-155.
[10]兰维娟,毛鹏军,杜东亮,等.基于径向基函数网络的变量施肥决策研究[J].安徽农业科学, 2007,35(21):6505-6507.
[11]冯沈萍等.运用人工神经网络预测一代粘虫发生程度的研究[J].微型电脑应用,1999(15):54-56
[12]马飞,许晓风,张夕林,等.神经网络预警系统及其在害虫预测中的应用.昆虫知识,2002,39(2):115-119..
[13] Drummond,S.Joshi,A.Sudduth,K.A.Application of neural networks: precision farming[A].Neural Networks Proceedings.1998,211-215.
[14] P.Larranaga,S.Moral.Probabilistic graphical models in artificial intelligence.Applied soft computing[J].2008, doi:10.1016/j.asoc.2008.01.003.
[15]张连文,郭海鹏.贝叶斯网引论[M].北京:科学出版社,2006.
[16] F.trai,A bayesian network for predicting yield response of winter wheat to fungicide programs[J].Computers and electronics in agriculture,1996:111-121.
[17] Kristian Kristensen,Ilse A.Rasumussen.The use of a Bayesian network in the design of a decision support system for growing malting barley without use of pestcides[J].Computers and electronics in agriculture.2002:197-217.
[18] Yonghua Qu,Jindi Wang,Huawei Wan,et al.A Bayesian network algorithm for retrieving characterization of land surface vegetation.Remote Sensing of Environmen[J].2008(112),613-622.
[19]于合龙,陈桂芬,董旭初.基于贝叶斯网的专家系统平台的开发与应用[J].黑龙江大学自然科学学报,2008,25(4):492-497.
[20]邓聚龙.农业系统灰色理论与方法[M].济南:山东科学技术出版社.1988.
[21]郭瑞林.作物灰色育种学[M].北京:中国农业科技出版社.1995.
[22]王占中,郭瑞林,齐光荣,等.基于Java的作物灰色育种智能决策系统研究.安徽农业科学, 2008,36(13):5669-5670,5675.
[23]陈桂芬,曹丽英,赵月玲.Fuzzy聚类分析在土壤养分分类中的应用[J].玉米科学,2008,16(2):118-121.
[24] Li yan,Shui Zhou,Li Feng, Li Hong-Yi.Delineation of site-specific management zones using fuzzy clustering analysis in acoastal saline land[J].Computers and electronics in agriculture,2007(56):174-186.
[25]陈彦,吕新.基于FCM的绿洲农田养分管理区研究[J].中国农业科学.2008,41(7):2016-2024.
[26]张璐,刘淑英,王平.基于模糊数学的农用地质量综合评价模型[J].广东农业科学,2010,2:207-2-9,215.
[27]王春平,张改生,张伟,等.应用模糊综合评判和灰色关联度分析评估小麦新品种[J].西北农林科技大学学报(自然科学版),2005,33(8):126-130)
[28]唐惠丽,周炼清,叶基瑶.基于模糊推理的农业病虫害专家系统[J].浙江农业学报,2009 ,21(5):506～509.
[29] Ambuel,J.R.,Colvin,et al. A fuzzy logic yield simulator for prescription farming[J].Transactions of ASAE.1994,37(6):1999-2009.
[30] Yang,C.C.,Prasher,et al.Development of an imag processing system and a fuzzy algorithm for site-specific herbicide applications.precision agriculture[J].Precision Agriculture,2003,4(1):5-18.
[31]唐启义,冯明光.DPS数据处理系统-实验设计、统计分析及数据挖掘[M].科学出版社,2007.1.
[32]全国农业技术推广服务中心.耕地地力指南评价指南[M].中国农业科学技术出版社,2006.
[33]赵春江.农业智能系统[M].北京:科学出版社.2009.1
[34]曹卫星.数字农作技术[M].北京:科学出版社.2008.8
[35] Y.cohen,A.Cohen,A.Hetzron,et al.Spatial decision support system for Medfly control in citrus[J]. computers and electronics in agriculture.2008,(62):107-117.
[36]陈蓉蓉.精确农作决策支持系统的研究[D].南京农业大学,2003.
[37]陈云坪,赵春江,王秀,等.基于知识模型与WebGIS的精准农业处方智能生成系统研究[J].中国农业科学,2007,40(6):1190-1197.
[38] Edson Murakami,Antomia M.Saraiva,Luiz C.M.Ribeiro Junior,etc.An infrastructure for the development of distributed service-oriented information systems for precision agriculture[J].computers and electronics in agriculture.2007(58):37-48.
[39] Allan Leck Jensen.Building a web-based information system for variety selection in field crops:objectives and results[J].Computers and Electronics in Agriculture.2001(32):195–211.
[40]高亮之.农业系统学基础[M].南京:江苏科学技术出版社,1993.
[41]罗锡文,张泰岭,洪添胜.“精细农业”技术体系及其应用[J].2001,32(2):103-106.
[42]赵春江,薛绪掌,王秀.精准农业技术体系的研究进展与展望[J].农业工程学报,2003,19(4):7-12.
[43]曹宏鑫,张春雷,金之庆,等.数字化栽培的框架与技术体系探讨[J].耕作与栽培,2005,(3):4-7,40.
[44]邬伦,刘瑜,等.地理信息系统-原理、方法与应用[M].北京:科学出版社,2001.
[45] N.A. Lorentzos , A.B. Sideridis, C.P. Yialouris ,et al.An integrated spatiotemporal system[J]. Computers and Electronics in Agriculture. 1999 (22) :233–242.
[46]王生生,刘大有,欧阳继红,等.数字农业时空信息管理平台[J].计算机应用研究,2007,24(5):163-165,185.
[47]张伟,数字农业空间信息管理平台开发研究[D].华东师范大学.2004.
[48]王娟,查良松.GIS与空间数据挖掘集成在农业中的应用[J].安徽农业科学,2008,36(29):12974-12975,12985.
[49]陈桂芬.面向精准农业的空间数据挖掘技术研究与应用[D].吉林大学.2009.
[50] MATTHEW L. ADAMS,U SIMON COOK, ROBERT CORNER.Managing Uncertainty in Site-Specific Management: What is the Best Model[J]. Precision Agriculture,2000,( 2): 39-54.
[51]刘思峰,党耀国,方志耕.灰色系统理论及其应用(第三版)[M].北京:科学出版社,2008.
[52] Dimitrios Moshou , Els Vrindts, Bart De Ketelaere. A neural network based plant classifier[J]. Computers and Electronics in Agriculture. 2001,(31): 5–16.
[53] J.A. Marchant, C.M. Onyango. Comparison of a Bayesian classifier with a multilayer feed-forward neural network using the example of plant/weed/soil discrimination[J]. Computers and Electronics in Agriculture. 2003 (39):3-22.
[54] Hansen L K,Salamon P.Neural network ensembles[C].IEEE Trans Pattern Analysis and Machine Intelligence,1990,12(10):993- 1001.
[55] Hansen L K, Salamon P. Neural network ensembles[C]. IEEE Trans PatternAnalysis and Machine Intelligence, 1990, 12 (10): 993- 1001.
[56] Sollich P, Krogh A. Learning with ensembles: How over-fitting can be useful[J]. In: TouretzkyD, MozerM , HasselmoMeds. Advances in Neural Information Processing Systems 8, Cambridge, MA: MIT Press, 1996. 190- 196.
[57] Schapire R E. The strength of weak learn ability[J]. Machine Learning, 1990, 5 (2): 197- 227.
[58] Breiman L. Bagging predictors[J]. Machine Learning, 1996, 24 (2) : 123- 140
[59] G. Peter Zhang: A neural network ensemble method with jittered training data for time series forecasting[J]. Inf. Sci. 177(23): 5329-5346 (2007).
[60] Zhi-Hua Zhou, Jianxin Wu, Wei Tang. Ensembling neural networks: Many could be better than all[J]. Artif. Intell.137(1-2): 239-263 (2002).
[61]吴建鑫,周志华,沈学华,等.一种选择性神经网络集成构造方法[J].计算机研究与发展.2000,37(9):1039-1044
[62] Zhi-Hua Zhou, Jianxin Wu, Yuan Jiang, Shifu Chen.Genetic Algorithm based Selective Neural Network Ensemble[C]. IJCAI 2001: 797-802.
[63] FU Qiang, HU Shang-xu, ZHAO Sheng-ying. Clustering-based selective neural network ensemble[J]. Journal of Zhejiang University Science,2005, 6A(5): 387-392.
[64]李凯,黄厚宽.一种基于聚类技术的选择性神经网络集成方法[J].计算机研究与发展.2005,42(4):594-598.
[65] Haixia Chen, Senmiao Yuan, Kai Jiang. Selective Neural Network Ensemble Based on Clustering[M]. ISNN (1) 2006: 545-550.
[66] Jian Lin, Bangzhu Zhu. A Novel Kernel Clustering Algorithm Based Selective Neural Network Ensemble Model for Economic Forecasting[C]. ISICA 2007:310-315.
[67] Jian Lin, Bangzhu Zhu. A novel neural network ensemble system for economic forecasting[C].ICIC 2007,CCIS 2, 2007: 1227-1233.
[68] Lean Yu, Shouyang Wang, Kin Keung Lai. A novel nonlinear ensemble forecasting model incorporating GLAR and ANN for foreign exchange rates[J]. Computers & OR 32: 2523-2541 (2005).
[69] Kin Keung Lai, Lean Yu, Shouyang Wang,Wei Huang. A Novel Nonlinear Neural Network Ensemble Model for Financial Time Series Forecasting[C]. International Conference on Computational Science (1) 2006: 790-793.
[70] Frey B J, Dueck D. Clustering by Passing Messages between Data Points[J]. Science, 2007, 315(5814): 972-976.
[71]高祥照,胡克林,郭炎,等.土壤养分与作物产量的空间变异特征与精确施肥[J].中国农业科学.2002:35(6):660-666.
[72]张书慧,马成林,等.变量施肥对玉米产量及土壤养分影响的实验[J].农业工程学报.2006,22(8):64-67.
[73]侯彦林,陈守伦.施肥模型研究综述[J].土壤通报.2004,35(4):493-501.
[74]王圣瑞,陈新平,高祥照等.“3414”肥料试验模型拟合的探讨[J].植物营养与肥料学报.2002,8(4) :409～413.
[75]刘光启.农业速查速算手册(上)--农业常用技术资料、土壤和肥料、农业设施卷[M].化学工业出版社.2008.
[76] Stuart Russell,Peter Norvig.Artificial Intelligence—A modern approach[M](影印版).北京:清华大学版社,2006.
[77] Joseph Giarratano,Gary Riley著,印鉴,刘星成,汤庸译.专家系统原理与编程[M].北京:机械工业出版社.2000.
[78]项利国,沈益民,何勇,等.粗糙集理论在粮食产量影响因素分析中的应用[J].浙江大学学报(农业与生命科学版),2003,29(3):343-346.
[79]于晓秋,葛家麒,刘长海.组合预测方法在黑龙江垦区大豆产量预测中的应用[J].数学的实践与认识,2007,37(24):27-32.
[80]王波,郭夜白,高来斌,等.最优加权预测在吉林省粮食产量预测中的应用[J].吉林农业大学学报. 2008,30(5):760-763.
[81]张谷丰等.基于WebGIS的农作物病虫害预警系统[J].农业工程学报,2007(23):176-181.
[82]段项锁,薛正平,杨星卫,等.精准农业土壤养分与水稻产量关系模型研究[J].中国生态农业学报.2003,11(1):62-65.
[83]薛正平,邓华,杨星卫,等.基于决策树和图层叠置的精准农业产量图分析方法[J].农业工程学报.2006,22(8):140-144.
[84] Georg Ru?, Rudolf Kruse1, Martin Schneider,et al. Data Mining with Neural Networks for Wheat Yield Prediction[J]. Advances in Data Mining.Medical Applications,E-Commerce, Marketing,and Theoretical Aspects. Springer Berlin/Heidelberg.2008.47-56.
[85] J. Cheng, R. Greiner, J. Kelly, D. Bell, W. R. Liu. Learning Bayesian networks from data: An information-theory based approach[J]. Artificial Intelligence, 2002(137): 43-90.
[86]陈争光,蔡德利.模糊物元分析法在耕地地力评价中的应用[J].安徽农学通报, 2008,14(11):55-56,233.
[87]田剑,胡月明,王长委,等.聚类支持下决策树模型在耕地评价中的应用[J].农业工程学报,2007,23 (12):58-63.
[88]周颜军,王双成,王辉.基于贝叶斯网络的分类器研究[J].东北师大学报自然科学版.2003,35(2):21-27.
[89]谢高地,陈沈斌.环境的空间连续变异与精准农业[M].北京:气象出版社.2005.7.
[90]宋新山,邓伟,张琳.MATLAB在环境科学中的应用[M].北京:化学工业出版社.2008.2.
[91]李翔,潘瑜春,赵春江,等.基于多年产量数据的精准逐农业管理分区提取与尺度效应评价[J].中国农业科学,2005,38(9):1825-1833.
[92]李艳,史舟,吴次方,等.基于多源数据的盐碱地精确农作管理分区研究[J].农业工程学报,2007,23(8):84-89.
[93]李翔,潘瑜春,马景宇,等.基于多种土壤养分的精准管理分区方法研究[J].土壤学报.2007,44(1):14-20.
[94] R.Khosla,D.Inman,D.G.Westfall,R.M.Reich,M.Frasier,M.Mzuku,B.Koch,A.Hornung.A synthesis of multi-disciplinary research in precision agriculture:site-specific management zones in the semi-arid western Great Plains of the USA[J].Precision agriculture.2008(9):85-100.
[95]郭龙珠,吕纯波,罗玉峰.应用因子分析法遴选水稻灌溉管理分区关键指标[J].农业系统科学与综合研究,2008,24(3):316-318.
[96]王子龙,付强,姜秋香.基于粒子群算法的土壤养分管理分区[J].农业工程学报,2008(10):80-84.
[97] Rodrigo A.Ortega,Oscar A. Santibanez.Determination of management zones in corn based on soil fertility[J]. Computers and electronics inagriculture. 2007(58):49-59.
[98]李艳双,曾珍香,张闽,等.主成分分析法在多指标综合评价方法中的应用[J].河北工业大学学报. 1999,28(1):94-97.
[99]陈贤,吴兴恩,杨德,等.主成分权重法在番茄果实商品性综合评价上的应用探讨[J].吉林农业科学.2008,33(4):49- 52.
[100]李柏年.加权模糊C-均值聚类[J].模糊系统与数学.2007,21(1):106-110.
[101]任军.气田开发方案模糊加权聚类分析[J].天然气工业.1993,14(2):35-37.
[102]谢季坚,刘承平.模糊数学方法及应用(第三版)[M].武汉:华中科技大学出版社.2005.
[103]李鸿吉.模糊数学基础及实用算法[M].北京:科学出版社.2005.
[104]李洁,高新波,焦李成.基于特征加权的模糊聚类新算法[J].电子学报.2006, 34( 1): 89-92.
[105] Sheng-sheng WANG, Da-you LIU , Spatio-temporal Database with Multi-granularities[C].Advances In Web-Age Information Management (WAIM 2004):PROCEEDINGS LECTURE NOTES IN COMPUTER SCIENCE 3129(LNCS):137-146 2004.
[106]邵维忠,杨芙清.面向对象的系统分析(第2版)[M].北京:清华大学出版社.2006.
[107]徐锋,陈暄.UML面向对象建模基础[M].北京:中国水利水电出版社.2006.
[108] Michael Negnevitsky.人工智能-智能系统指南[M].北京:机械工业出版社.2005.