论文部分内容阅读
介绍一种区间小波的构造方法.并将区间小波与神经网络相结合,提出一种用于信号分类的分类区间小波网络,利用它解决小波网络的基底空间与被学习信号所属空间不匹配的问题.在分类区间小波网络模型中引入模拟退火策略,并采用自适应变学习系数训练网络.实验结果表明,将分类区间小波网络应用于雷达目标识别,可以减少神经元数目,提高网络收敛速度,并能较好解决高维学习的“维数灾难”问题,获得较好的分类效果.
This paper introduces a construction method of interval wavelet, and combines interval wavelet with neural network to propose a classification interval wavelet network for signal classification, which is used to solve the problem that the space between the base space of wavelet network and the space to be learned belongs The simulated annealing strategy is introduced in the classification interval wavelet network model and the adaptive learning coefficient is used to train the network.The experimental results show that the classification interval wavelet network can be applied to radar target recognition to reduce the number of neurons and improve the network convergence rate Can better solve the problem of “dimensionality disaster” of high-dimensional learning and get better classification results.