论文部分内容阅读
针对现有基于误差反向传播算法的多层感知器神经网络分类器在信号识别中存在收敛速度缓慢、出现假饱和现象等问题,采用蜂群算法提取信号的联合特征模块,提出快速支持、超级自适应误差反向传播、共轭梯度3种不同算法分别应用于多层感知器神经网络分类器,实现对通信信号的自动识别。所提算法和误差反向传播算法相比有更高的识别率。仿真结果表明,所提算法能够克服误差反向传播算法的缺陷,在隐藏层神经元仅为20个、信噪比为4dB条件下,3种算法的识别率均高于95%,且系统易于实现,在信号识别中具有广泛的应用前景。
In order to solve the problem of slow convergence rate and false saturation in signal recognition, existing multi-layer perceptron neural network classifiers based on error backpropagation algorithm use the bee colony algorithm to extract the joint feature modules of signals and propose fast support, super Adaptive error backpropagation, conjugate gradient three kinds of different algorithms were applied to multi-layer perceptron neural network classifier, to achieve the automatic identification of communication signals. The proposed algorithm has a higher recognition rate than the error backpropagation algorithm. The simulation results show that the proposed algorithm can overcome the defect of error back propagation algorithm. The recognition rate of the three algorithms is higher than 95% with only 20 hidden neurons and SNR of 4dB, and the system is easy Achieve, have a wide range of application prospects in signal recognition.