微波干燥过程的神经网络模型  

Modeling of Microwave Drying by Artificial Neural Networks

作　　者：朱代根[1] 

机构地区：[1]西南林学院交通机械与土木工程学院,昆明650224

出　　处：《农机化研究》2010年第6期94-96,共3页Journal of Agricultural Mechanization Research

摘　　要：研究了微波干燥过程的神经网络模型。建立了基于Levenberg—Marquardt优化算法的3层BP神经网络,网络输入为微波功率和辐射时间,输出为干基湿含量。使用线性回归方程、均方差和相关系数来评定所建网络的性能。结果表明,当隐层激发函数和神经元个数分别为tansig和25时,所得模型可以同时满足训练和测试的精度要求。The main objective of this work was to develop artificial neural network （ANN） model for microwave drying. One-hidden-layer feed-forward ANN models with Levenberg-Marquardt algorithm were trained and validated with experimental data. Inputs for ANN models were microwave power and drying time. Output from ANN models was moisture content at a given drying time. The model was evaluated by regression line,standard deviation and correlation coefficient. Based on our results,the best network was composed of 25 hidden nodes and used a tangent hyperbolic（tansig） transfer function in the first layer.

关 键 词：微波干燥 神经网络 模型 

分 类 号：S226.6[农业科学—农业机械化工程]