人工神经网络模型及其在农业和生态学研究中的应用

doi:10.17521/cjpe.2005.0114

摘要/Abstract

摘要：

对于一些复杂的农业生态系统,人们对其生态过程了解较少,且这些系统的不确定性和模糊性较大,用传统的方法难以模拟这些系统的行为,神经网络模型因为能较精确地模拟这些系统的行为,而引起生态学者们的广泛兴趣。该文着重介绍了误差逆传神经网络模型的结构、算法及其在农业和生态学中的应用研究。误差逆传神经网络模型一般采用三层神经网络模型结构,三层的神经网络模型能模拟任意复杂程度的连续函数,而且因为它的结构小而不容易产生与训练数据的过度吻合。误差逆传神经网络模型算法的主要特征是:利用当前的输入误差对权值进行调整。在生态学和农业研究中,误差逆传神经网络模型通常作为非线性函数模拟器用于预测作物产量、生物生产量、生物与环境之间的关系等。已有的研究表明:误差逆传神经网络模型的模拟精度要远远高于多元线性方程,类似于非线性方程,而在样本量足够的情况下,有一定的外推能力。但是误差逆传神经网络模型需要大量的样本量来保证所求取参数的可靠性,但这在实际研究中很难做到,因而限制了误差逆传神经网络模型的应用。近年来人们提出了强制训练停止、复合模型等多种技术来提高误差逆传神经网络模型的外推能力,也提出了Garson算法、敏感性分析以及随机化检验等技术对误差逆传神经网络模型的机理进行解释。误差逆传神经网络模型的真正优势在于模拟人们了解较少或不确定性和模糊性较大系统的行为,这些是传统模型所无法实现的,因而是对传统机理模型的重要补充。

关键词: 人工神经网络, 误差逆传, 农业及生态系统, 机理模型, 非线性模型

Abstract:

Artificial neural network appeals to many ecologists because of the complexity of agroecological systems and its precise predicting ability to simulate vaguely understood and highly uncertain ecosystems. We focus on the introduction of the structure, algorithm and the applications for back-propagation artificial neural network (BPN) in this paper. People usually adopt BPN of a three-layer structure that can approximate functions of any complexity and is not easy to overfit because of its simple structure. The main principle of back-propagation artificial neural network is to adjust weights according to the errors of input neurons. In agroecological research, neural networks are always employed to predict crop yield, biomass yield, relationships between organisms and environmental factors and so on. Previous studies indicate that neural networks greatly outperform linear models, while the accuracy of the results produced by neural networks are very similar to that of algorithmic models. Moreover, neural networks can extrapolate to some degree when enough training data are provided. However, neural networks require a large number of samples to guarantee the robustness of its parameterization, which seems unrealistic for complicated networks to collect such large data sets from crop growth experiments involving countless small plot trials over multiple site-years. The requirement of large amount of training data has hindered the application of neural networks. Some techniques, such as early stopping, jittering and metamodels, have been advanced to induce generalization of neural networks, while techniques such as Garson's algorithm, sensitivity analysis and randomization test are advanced to explain the mechanisms of neural networks. The advantage of neural network lies on its ability to precisely simulate the vaguely understood and uncertain ecosystems, which cannot be realized by traditional approaches. As a nonlinear approximator, artificial neural network is an important tool complementary to comprehensive models.

Key words: Artificial neural networks, Back-propagation, Agro-ecological system, Comprehensive model, Nonlinear model

米湘成, 马克平, 邹应斌. 人工神经网络模型及其在农业和生态学研究中的应用. 植物生态学报, 2005, 29(5): 863-870. DOI: 10.17521/cjpe.2005.0114

MI Xiang-Cheng, MA Ke-Ping, ZOU Ying-Bin. ARTIFICIAL NEURAL NETWORK AND ITS APPLICATION IN AGRICULTURAL AND ECOLOGICAL RESEARCH. Chinese Journal of Plant Ecology, 2005, 29(5): 863-870. DOI: 10.17521/cjpe.2005.0114

导出引用管理器 EndNote|Ris|BibTeX

链接本文: https://www.plant-ecology.com/CN/10.17521/cjpe.2005.0114

https://www.plant-ecology.com/CN/Y2005/V29/I5/863

图/表 3

图1 神经网络结构示意图,图中是一个3层的前馈型神经网络,由输入层、隐含层和输出层3层构成,X为输入向量,Y为输出向量(Lek & Guégan, 1999)

Fig.1 Schematic illustration of a three-layered feed-forward neural network, with one input layer, one hidden layer and one output layer. X is the input vector, Y is the output vector (Lek & Guégan, 1999)

图2 8个环境变量与鱼的物种丰富度关系的神经网络机制图 (Olden & Jackson, 2002)

Fig.2 NID for neural network modeling fish species richness as a function of eight habitat variables (Olden & Jackson, 2002)

表1 1个BPN网络模型神经元之间的连接权值

Table 1 Axon connection weights of one neural networks

权值\输入神经元 Weights/Input neuron	输入变量 1 Input neuron 1	输入变量2 Input neuron 2	输入变量3 Input neuron 3	输出神经元 Output neuron
隐含神经元A Hidden neuron A	0.458 8	0.965 3	1.966 8	1.650 0
隐含神经元B Hidden neuron B	0.975 6	1.327 7	0.616 3	0.864 5
隐含神经元C Hidden neuron C	0.387 1	0.425 7	0.472 4	1.634 8
隐含神经元D Hidden neuron D	0.028 2	4.588 3	6.694 3	1.086 7

参考文献 29

[1]	Broner I, Comstock CR (1997). Combining expert systems and neural networks for learning site-specific conditions. Computers and Electronics in Agriculture, 19,37-53.
[2]	Corne SA, Carver SJ, Kunin WE, Lennon JJ, van Hees WWS (2004). Predicting forest attributes in southeast Alaska using artificial neural networks. Forest Science, 50,59-276.
[3]	Demuth H, Beale M (2001). Neural Network Toolbox for Use with MATLAB. Natick, the MathWorks, Inc.
[4]	Garson GD (1991). Interpreting neural-network connection weights. Artificial Intelligence and Expert System, 6,47-51.
[5]	Goh AT (1995). Back-propagation neural networks for modelling complex systems. Artificial Intelligence in Engineering, 9,143-151.
[6]	James FC, McCulloch CE (1990). Multivariate analysis in ecology and systematics: pannacea or Pandora's box? Annual Review of Ecology and Systematics, 21,129-196.
[7]	Kavdir I (2004). Discrimination of sunflower, weed and soil by artificial neural networks. Computers and Electronics in Agriculture, 44,153-160.
[8]	Lee JHW, Huang Y, Dickman M, Jayawardena AW (2003). Neural network modelling of coastal algal blooms. Ecological Modelling, 159,179-201.
[9]	Lek S, Beland A, Dimopoulos I, Lauga J, Moreau J (1995). Improved estimation, using networks of the food consumption of fish populations. Marine Freshwater Research, 46,1229-1236.
[10]	Lek S, Delacoste M, Baran P, Dimopoulos I, Lauga J, Aulagnier S (1996). Application of neural networks to modelling nonlinear relationships in ecology. Ecological Modelling, 90,39-52.
[11]	Lek S, Guégan JF (1999). Artificial neural networks as a tool in ecological modeling: an introduction. Ecological Modelling, 120,65-73.
[12]	Mi XC, Zou YB, Wei W, Ma KP (2005). Testing the generalization of artificial neural networks with cross validation and independent validation in modelling rice tillering dynamics. Ecological Modelling, 181,493-508.
[13]	Moisen GG, Frescino TS (2002). Comparing five modelling techniques for predicting forest characteristics. Ecological Modelling, 157,209-225.
[14]	Olden JD, Jackson DA (2002). Illuminating the "black box": a randomization approach for understanding variable contributions in artificial neural networks. Ecological Modelling, 154,135-150.
[15]	Olden JD, Joy MK, Death RG (2004). An accurate comparison of methods for quantifying variable importance in artificial neural networks using simulated data. Ecological Modelling, 178,389-397.
[16]	Özesmi SL, Özesmi U (1999). An artificial neural network approach to spatial habitat modeling with interspecific interaction. Ecological Modelling, 116,15-31.
[17]	Park YS, Cereghino R, Compin A, Lek S (2003). Applications of artificial neural networks for patterning and predicting aquatic insect species richness in running waters. Ecological Modelling, 160,265-280.
[18]	Ryan M, Muller C, Di HJ, Cameron KC (2004). The use of artificial neural networks (ANNs) to simulate N₂O emissions from a temperate grassland ecosystem. Ecological Modelling, 175,189-194.
[19]	Scardi M (2001). Advances in neural network modelling of phytoplankton primary production. Ecological Modelling, 146,33-45.
[20]	Schultz A, Wieland R (1997). The use of neural networks in agroecological modelling. Computer and Electronics in Agriculture, 18,73-90.
[21]	Schultz A, Wieland R, Lutze G (2000). Neural networks in agroecological modelling -stylish application or helpful tool? Computer and Electronics in Agriculture, 29,73-97.
[22]	Segurado P, Araújo MB (2004). An evaluation of methods for modelling species distributions. Journal of Biogeography, 31,1555-1568.
[23]	Starrett SK, Najjar Y, Adams G (1998). Modeling pesticide leaching from golf courses using artificial neural networks. Communications in Soil Science and Plant Analysis, 29,3093-3106.
[24]	Tamari S, Wosten JHM, Ruiz SJC (1996). Testing an artificial neural network for predicting soil hydraulic conductivity. Soil Science Society of America Journal, 60,1732-1741.
[25]	ter Braak CJF (1986). Canonical correspondence analysis: a new eigenvector technique for multivariate direct gradient analysis. Ecology, 67,1167-1179.
[26]	van Wijk MT, Bouten W, Verstraten JM (2002). Comparison of different modelling strategies for simulating gas exchange of a Douglas-fir forest. Ecological Modelling, 158,63-81.
[27]	Viotti P, Liuti G, Di Genova P (2002). Atmospheric urban pollution: applications of an artificial neural network (ANN) to the city of Perugia. Ecological Modelling, 148,27-46.
[28]	Wang YN (王耀南) (1999). Computational Intelligent Information Processing: Technology and Applications (计算机智能信息处理技术及其应用). Hunan University Press, Changsha,30-155. (in Chinese)
[29]	Wolf ED, Francl LJ (1998). Empirical infection period models for tan spot of wheat. Canadian Journal of Plant Pathology, 20,394-395.