基于主成分分析与模糊BP方法的藻类繁殖状态预测

摘要/Abstract

摘要： 将主成分分析（PCA）与模糊反向传播（BP）网络建模方法相融合，提出了PCA-模糊BP方法并用于藻类繁殖状态的预测，建立了叶绿素a含量的预测模型.采用PCA对各类采集数据进行预处理，并将PCA所得各理化因子作为模糊BP网络的输入变量，叶绿素a的含量作为模糊BP网络的输出变量，经过学习训练，获得藻类繁殖状态的预测模型.结果表明，PCA-模糊BP方法降低了各类输入样本数据之间的相关性和模型系统的维数，加快了模糊BP网络的收敛速度，其与典型BP神经网络模型相比，具有更快的计算速度和更高的预测精度，能够较好地预测海洋藻类繁殖的生长状况.

关键词: 主成分分析, 模糊BP网络模型, 叶绿素a, 状态预测

Abstract: Principal component analysis (PCA) method combined with fuzzy back propagation (BP) network, called as PCA-fuzzy BP method, was proposed to predict the state of algae reproduction, and the prediction model of chlorophyll-a concentration was established. PCA method was used to preprocess various acquisition data, it reduces the dimensionality of the input data of the system. The physicalchemical factors produced by PCA processing can be regarded as the input variables of the fuzzy BP network, the concentration of chlorophylla can be regarded as the output of fuzzy BP network, and the state prediction model of algae reproduction can be obtained by learning and training for this network. The experimental result indicates that PCA method reduces the correlation for the factors of input sample data, also reduces the dimension of the model system, accelerates the speed of modelling convergence for fuzzy BP network. The PCA-fuzzy BP algorithm has a faster speed of calculation and more accuracy of prediction than the typical BP network. This kind of model can give a better state prediction for algae reproduction.

Key words: principal component analysis (PCA), fuzzy back propagation (BP) network model, chlorophyll-a, state prediction

中图分类号:

X 55

张颖1, 李彩娟1, 邵惠鹤2. 基于主成分分析与模糊BP方法的藻类繁殖状态预测[J]. 上海交通大学学报（自然版）, 2012, 46(05): 785-789.

ZHANG Ying-1, LI Cai-Juan-1, SHAO Hui-He-2. State Prediction of Algae Reproduction Based on PCA-Fuzzy BP Method[J]. Journal of Shanghai Jiaotong University, 2012, 46(05): 785-789.

参考文献

［1］Lee J H W, Huang Y, Dickman M, et al. Neural network modelling of coastal algal blooms ［J］. Ecological Modelling, 2003, 159 (2/3): 179/201.

［2］Melesse A M, Krishnaswamy J, Zhang K Q. Modeling coastal eutrophication at Florida bay using neural networks ［J］. Journal of Coastal Research, 2008, 24(2)：190196.

［3］Gonzalez V L, Evangelos S, Palenzuela T. Neural network estimation of chlorophyll a from MERIS full resolution data for the coastal waters of Galician rias (NW Spain) ［J］. Remote Sensing of Environment, 2011, 115 (2): 524535.

［4］Naz M, Uyanik S, Yesilnacar M I, et al. Sidebyside comparison of horizontal subsurface flow and free water surface flow constructed wetlands and artificial neural network (ANN) modelling approach ［J］. Ecological Engineering, 2009, 35 (8): 12551263.

［5］Hou G X, Li H B, Recknage F. Modeling phytoplankton dynamics in the river darling (Australia) using the radial basis function neural network ［J］. Journal of Freshwater Ecology, 2006, 21 (4): 639647.

［6］Talib A, Recknagel F, Cao H. Forecasting and explanation of algal dynamics in two shallow lakes by recurrent artificial neural network and hybrid evolutionary algorithm［J］. Mathematics and Computers in Simulation, 2008, 78 (2): 424434.

［7］Mozejko J, Gniot R. Application of neural networks for the prediction of total phosphorus concentrations in surface waters ［J］. Polish J of Environ Stud, 2008, 17 (3): 363368.

［8］Jeong KwangSeuk, Kim DongKyun, Joo GeaJae. River phytoplankton prediction model by artificial neural network: Model performance and selection of input variables to predict timeseries phytoplankton proliferations in a regulated river system［J］. Ecological Informatics, 2006, 1 (3): 235245.

［9］Ye L, Cai Q H. Forecasting daily chlorophyll a concentration during the spring phytopiankton bloom period in Xiangxi bay of the threegorges reservoir by means of a recurrent artificial neural network ［J］. Journal of Freshwater Ecology, 2009, 24 (4): 609617.

［10］王立新. 自适应模糊系统与控制设计与稳定性分析［M］. 北京：国防工业出版社, 1995: 3556.

［11］刘载文,吕思颖,王小艺,等. 河湖水华预测方法研究［J］. 水资源保护, 2008, 24 (5):4247.

LIU Zaiwen, L Siying, WANG Xiaoyi, et al. Forecast methods for algal bloom in rivers and lakes［J］. Water Resources Protection, 2008, 24 (5):4247.

［12］王洪礼, 葛根, 李悦雷. 基于模糊神经网络(FNN)的赤潮预警预测研究［J］. 海洋通报, 2006, 25 (4): 3641.

WANG Hongli, GE gen, LI Yuelei. Research on the prediction of red tide based on the fuzzy neural network ［J］. Marine Science Bulletin, 2006, 25 (4): 3641.

[1]	来颜博, 阎高伟, 程兰, 陈泽华. 基于动态独立成分分析和动态主成分分析的测地线流式核无监督回归模型[J]. 上海交通大学学报, 2020, 54(12): 1269-1277.
[2]	高雅1，王占永1，2，路庆昌1，彭仲仁1. 城市高架路沿侧细颗粒物的垂直分布特征研究[J]. 上海交通大学学报（自然版）, 2018, 52(6): 650-657.
[3]	吴军a, b，黎国强a，吴超勇a，程一伟c，邓超c. 数据驱动的滚动轴承性能衰退状态监测方法[J]. 上海交通大学学报（自然版）, 2018, 52(5): 538-544.
[4]	高筱婷，杨东升. 基于数据驱动的智能电器运行状态监测方法 [J]. 上海交通大学学报（自然版）, 2017, 51(9): 1104-1110.
[5]	张伟，许爱强，高明哲. 一种基于积累一致性测量的在线状态预测算法[J]. 上海交通大学学报（自然版）, 2017, 51(11): 1391-1398.
[6]	余焕伟，叶震，张志芬，陈华斌，陈善本. Al-Mg合金钨极气体保护焊接电弧的光谱特征提取方法[J]. 上海交通大学学报（自然版）, 2013, 47(11): 1655-1660.
[7]	杨漪1，姚晓栋1，杨建国1，张余升2，袁峰2. 基于主成分分析与BP神经网络相结合的机床主轴热漂移误差建模[J]. 上海交通大学学报（自然版）, 2013, 47(05): 750-753.
[8]	李华怡, 刘炜. 基于主成分分析的磨矿旋流器机理模型[J]. 上海交通大学学报（自然版）, 2012, 46(12): 1936-1939.
[9]	邵小龙,朱将伟，李云飞. 批量甜玉米低场核磁共振数据的统计分析 [J]. 上海交通大学学报（自然版）, 2011, 45(01): 144-0148.