摘要
人工神经网络是人工构造的模拟人脑功能而构建的一种网络,其中BP学习算法是一种单向传播的多层前向网络。利用MATLAB神经网络工具箱,设计BP网络的训练步骤,从园艺试验田观测的数据中抽取部分数据作为样本预测西瓜仁重,把影响西瓜种仁重的4个主要因素作为网络的输入向量,西瓜种仁重为网络的输出向量,根据需要调用相关程序,从而避免了编写大量的算法程序,训练结果表明,预测结果与实验结果吻合。
Artificial neural network is a man-made network to simulate the human brain structure and function, in which BP learning algorithm is a one-way transmission of multi-layer to the network. MATLAB toolbox neural net- work is applied to design BP network of training steps, taken part of watermelon seeds weight of the observation data in the experimental field as a sample of prediction data, four main factors that affect watermelon seeds weight as the network input vector,while taken watermelon seeds weight as output vector for the network, so as to avoid prepara- tion of a large number of algorithms procedures. The results showed that the predicted results accord with the experimental results.
出处
《安徽农业大学学报》
CAS
CSCD
北大核心
2008年第4期593-596,共4页
Journal of Anhui Agricultural University
基金
安徽省教育厅项目(KJ2007B158)资助
关键词
BP神经网络
预测
西瓜仁重
BP neural network
prediction
weight of watermelon seeds
