摘要
将混合神经网络 (PFNN FG)技术应用于植物病害预测 ,其输入矢量含模糊分量 ,遗传算法优化配置各参数。变形 Sigm oid函数用于不同的隐含层 ,构成参数化神经网络。网络的输入层引入模糊集合理论 ,使网络能处理语义变量。将 PFNN FG和其他神经网络 (如前向神经网络、径向基神经网络等 )用于大豆基准问题进行分析比较 ,结果是 PFNN FG在精度和训练速度上优于其他网络。将 PFNN FG和前向神经网络用于 2组黄瓜霜霉病数据 ,前者测试组的均方根误差小于后者。
This paper proposed a hybrid neural network based on parametric feedforward neural networks with fuzzy inputs configured by a genetic algorithm (PFNN _FG). A variant Sigmoid function was used at various hidden layers. The fuzzy set theory was employed at the input layer to make the processing of linguistic variables possible. The parameters of the variant Sigmoid function and the fuzzy parameters were configured by a genetic algorithm. A comparative analysis between PFNN _ FG and other neural networks on benchmark problems shows that PFNN _ FG is comparable with the other networks in terms of accuracy of the obtained results, but it is much faster.
出处
《农业机械学报》
EI
CAS
CSCD
北大核心
2004年第6期110-114,共5页
Transactions of the Chinese Society for Agricultural Machinery
关键词
植物
病虫害防治
参数化神经网络
模糊集
遗传算法
Plants, Techniques of pest control, Parametric neural network, Fuzzy set, Genetic algorithm
