改进粒子群优化超限学习机的调制信号识别  

Modulation signal recognition based on improved particle swarm optimization extreme learning machine

作　　者：梁猛 赵贝 LIANG Meng;ZHAO Bei(School of Electronic Engineering,Xi’an University of Posts and Telecommunications,Xi’an 710121,China)

机构地区：[1]西安邮电大学电子工程学院,陕西西安710121

出　　处：《西安邮电大学学报》2023年第2期57-64,共8页Journal of Xi’an University of Posts and Telecommunications

摘　　要：针对现有算法在信噪比较小条件下对调制信号识别精度较差的问题,提出一种基于改进粒子群优化超限学习机的调制信号识别算法。以调制信号的高阶累积量为基础构建4种特征参数,并根据多进制相移键控(Multiple Phase Shift Keying,MPSK)信号的相位特性引入归一化瞬时相位平均值和递归归一化瞬时相位平均值特征两个特征参数,构建调制信号的6种特征参数的数据集。利用粒子群优化(Particle Swarm Optimization,PSO)算法优化超限学习机(Extreme Learning Machine,ELM)算法的神经网络结构,动态调整PSO算法中的惯性权重,以提升算法的识别性能。采用所提算法对7种调制信号进行识别,实验结果表明,当信噪比大于2 dB时,所提算法对7种调制信号的识别正确率均达到100%。与相关算法相比,所提算法的识别效果较佳且更具有稳定性。Aiming at the problem that the existing algorithms have poor recognition accuracy of modulation signals under the condition of low signal-to-noise ratio(SNR),a recognition algorithm based on improved particle swarm optimization extreme learning machine(IPSO-ELM)is proposed.Four characteristic parameters are constructed based on the high-order cumulants of the modulated signal,and according to the phase characteristics of the multiple phase shift keying(MPSK)signal,the normalized instantaneous phase average and the recursive normalized instantaneous phase average are introduced to construct the data set of the six characteristic parameters of the modulated signal.The particle swarm optimization(PSO)algorithm is used to optimize the neural network structure of the ELM algorithm,and the inertia weight in the PSO algorithm is dynamically adjusted to improve the recognition performance of the algorithm.The experimental results show that when the SNR is greater than 2 dB,the identification accuracy of the proposed algorithm for all seven modulation signals reaches 100%.The proposed algorithm has a better recognition effect and more stability than the related algorithms.

关 键 词：改进粒子群优化超限学习机 调制信号识别 高阶累积量 归一化瞬时相位平均值 递归归一化瞬时相位平均值 

分 类 号：TN911.7[电子电信—通信与信息系统]