摘要
针对木材含水率的检测精度受到多种环境因子(温度、平衡含水率、风速等)的影响,提出了一种基于函数联接型神经网络FLANN的数据融合方法以消除环境温度对木材含水率检测精度的扰动。FLANN能利用函数扩展替代多层感知器(MLP)的隐含层,将输入信号空间维度下的线性不可分类的问题,扩展至较高信号空间维度的超平面上,简化了MLP的架构。与MLP相比,FLANN具有结构简单、收敛速度快和计算量小的特点。仿真结果及实验验证表明:基于FLANN的数据融合方法,能有效消除环境因子的扰动,并可实现木材含水率的稳定、实时、高精度检测。
The measurement precise of wood moisture content（WMC） is disturbed by many ambient factors such as temperature,equilibrium moisture content and wind speed;therefore a new testing method for wood moisture content（WMC） was put forward based on Functional Link Artificial Neural Networks（FLANN） in order to eliminate the disturbances caused by ambient factors.In the FLANN,functional expansion could substitute the hidden layer of multilayer perceptron（MLP）,and the dimension of the input signal space was enhanced by polynomials.Compared with MLP,FLANN exhibits a much simpler structure,faster convergence and less computational complexity.The calibration tests and simulation studies show that the FLANN-based data fusion technique can effectively eliminate the disturbances caused by ambient factors,and steady,real-time,high-accuracy measurement of wood moisture content can be realized.
出处
《东北林业大学学报》
CAS
CSCD
北大核心
2010年第12期65-68,共4页
Journal of Northeast Forestry University
基金
国家自然科学基金项目(30771678)
关键词
函数联接型神经网络
数据融合
多层感知器
木材含水率
Functional link artificial neural networks
Data fusion
Multilayer perceptron
Wood moisture content
