一种引力搜索优化神经网络的调制识别算法

A modulation recognition method based on gravitational search optimizing neural network

为了提高低信噪比条件下调制信号识别率,提出一种基于引力搜索算法优化神经网络的数字调制识别算法。首先通过瞬时特征、高阶累积量和小波变换提取信号的6个特征参数。然后基于Tent映射初始化神经网络的权值和阈值,降低局部收敛的概率;当陷入局部最优时,对精英粒子进行柯西扰动,增加全局搜索能力;通过自适应改变引力衰减因子代替原有固定取值,从而提高引力搜索算法的收敛速度。最后利用混合引力搜索算法优化神经网络,将特征参数输入优化后的神经网络分类器对信号进行分类识别。仿真结果表明,该识别算法在信噪比为-1dB时,信号的整体识别率达到95%以上。

In order to improve the recognition rate of modulation signal at low Signal-to-Noise Ratio(SNR),a digital modulation recognition algorithm based on gravitational search algorithm optimize the neural network is proposed.In this algorithm,the six features of the signal are firstly extracted by using the time domain features,high-order cumulant and wavelet transform.Using the Tent chaotic map to initialize the population,the probability that the particles fall into the local optimum is then reduced.The Cauchy perturbation of the elite particles is carried out when local optimum is falled into,the global search ability of the algorithm is then enhanced.Secondly,by adaptively changing the attenuation factor gravitation fixed value instead of the original,the convergence speed of the algorithm is improved and finally the improved neural network model is used to identify signals.Simulation results show that the average recognition rate is more than 95%with SNR at-1 dB,which proves the validity of the method.