本文以小波理论的多分辨率分析为基础,分析了各种小波网络的频率特性,指出了由尺度函数和小波函数组成的多分辨率小波网络的优点.针对多分辨率小波网络,本文提出了一种在线辨识算法.该算法,当增加样本和新神经元时,在线修正网络权值,同时通过正交化在线优选神经元,达到优化网络结构和网络权值的目的,训练速度快,辨识精度高,仿真结果表明了该算法的有效性. Based on the multi-resolution analysis of wavelet theory, this paper analyzes the frequency characteristics of various wavelet networks and points out the advantages of multi-resolution wavelet networks composed of scale functions and wavelet functions. For multiresolution wavelet networks, this paper proposes An online recognition algorithm is proposed in this algorithm. When the sample and the new neurons are added, the network weights are modified online and the neurons are optimized online by orthogonalization so as to optimize the network structure and network weights. The training speed is fast, The precision is high. Simulation results show the effectiveness of the algorithm.