摘要
以单隐层前向神经网络为基础,对纤维板热压传热传质过程中的相对湿度进行参数识别。分析纤维板热压传热传质过程中混合气体相对湿度这一影响因子,利用多项式神经网络逼近定理及筛减原理,给出识别这一影响因子的专门算法及仿真结果。仿真结果与Engelhardt(1979)给出的榉木试件的等湿线比较可以看出,实验值与实测值曲线十分接近,表明该方法是行之有效的。本结果对纤维板热压过程中的产品质量控制有一定的指导作用。
The main work is to identify the parameters of relative humidity in the process of heat and mass transfer for fiberboard hot-pressing by using a single hidden layer forward neural network. The impact factors of the relative humidity in the process of fiberboard hot-pressing are studied. A specific algorithm to identify the impact factors is given by using the screen cut principle and polynomial neural network approximation theorem. Compared the simulation results with the constant humidity curve of beech specimens given by Engelhardt ( 1979 ), it is shown that the experimental values and the measured values have more close curves so the method is effective. The obtained results have a certain guiding role for the product quality control in the process of fiberboard hot-pressing.
出处
《黑龙江大学自然科学学报》
CAS
北大核心
2017年第1期37-41,共5页
Journal of Natural Science of Heilongjiang University
基金
黑龙江省自然科学基金资助项目(A2015016)
关键词
纤维板
热压
神经网络
参数识别
fiberboard
hot-pressing
neural network
parameters identification