摘要
基于B样条小波基具有很强的自适应数据或函数变化的能力,本文构造了二类B样条小波神经网络,使之逼近于非线性多输入多输出系统,讨论了这些网络的结构形式和结构特点,并给出了其相应的算法.最后将此理论应用于复合材料的应变损伤位置的诊断.
Seeing that B-spline wavelets have the ability to smooth datum and adapt to the changes of function, we have developed two classes of feedforward B-spline wavelet neural networks for approximating arbitrary nonlinear systems. Different network structures and energy functions are necessary and are given for representation and classification. Algorithm of backpropagation type are proposed for B-spline wavelet network training. Applying the concepts in the paper to detect the damage position in the smart composite structures shows that B-spline wavelet neural networks are more accurate and faster in the learning speed than the BP network.
出处
《模式识别与人工智能》
EI
CSCD
北大核心
1996年第3期228-233,共6页
Pattern Recognition and Artificial Intelligence
基金
国家自然科学基金
航空科学基金
关键词
B样条
函数逼近
小波
神经网络
B-Spline Wavelet, Neural Network, Classification, Smart Composite Structures, Damage Detect.