Study on Pests Forecasting Using the Method of Neural Network Based on Fuzzy Clustering  

基于模糊聚类的神经网络虫情预测

作　　者：韦艳玲[1] 

机构地区：[1]柳州职业技术学院信息工程系,广西柳州545006

出　　处：《Agricultural Science & Technology》2009年第4期159-163,共5页农业科学与技术（英文版）

基　　金：Supported by Guangxi Science Research and Technology Explora-tion Plan Project(0815001-10)~~

摘　　要：Aimed to the characters of pests forecast such as fuzziness, correlation, nonlinear and real-time as well as decline of generalization capacity of neural network in prediction with few observations, a method of pests forecasting using the method of neural network based on fuzzy clustering was proposed in this experiment. The simulation results demonstrated that the method was simple and practical and could forecast pests fast and accurately, particularly, the method could obtain good results with few samples and samples correlation.1基于模糊聚类的神经网络预测首先采用模糊聚类对所有样本进行预处理，再把去噪后的数据作为神经网络的输人数据进行训练和预测。1．1基于模糊聚类的神经网络结构3层BP神经网络具有令人满意的对连续映射的逼近能力，可以满足预测的要求，因此，采用3层BP神经网络作为研究模型。3层BP神经网络由输入层、隐含层和输出层组成。

关 键 词：Neural network Fuzzy clustering PEST Forecasting 

分 类 号：S431[农业科学—农业昆虫与害虫防治]