Abstract: Currently, many recognition methods of digital modulation signals occur, which have different recognition effects. In order to improve recognition performance of the signals under different Signal-to-Noise Ratio（SNR）, a method based on Modified Particle Swarm Optimization（MPSO） algorithm optimizing BP is proposed. Six instantaneous feature parameters are extracted, among which the [Rσa] is an improvement and the [Rσp] is an analogy of the [Rσa]. The neighbor information of particles are referred to modify Particle Swarm Optimization（PSO） algorithm optimizing the weights and thresholds of BP. Simulation result shows that the method has a remarkable performance because the probability to recognize seven kinds of signals is over 86%.

Key words: digital modulation signals, instantaneous feature parameters, neighbor information, Particle Swarm Optimization（PSO） algorithm, Back Propagation（BP）

摘要： 数字调制信号的识别方法有很多，其识别效果不尽相同。为了提高数字调制信号在不同信噪比（Signal-to-Noise Ratio，SNR）下的识别性能，提出了一种基于改进粒子群（Modified Particle Swarm Optimization，MPSO）算法优化BP网络的识别方法。针对七种常见的数字调制信号，提取了六个瞬时特征参数，其中[Rσa]参数是改进得到的，同理类推得到[Rσp]。为了在保持基本粒子群（Particle Swarm Optimization，PSO）算法优点的基础上进一步提高算法的性能，增加了对粒子邻域信息的参考，再用MPSO算法优化BP网络的权值和阈值。从仿真实验可以看出，应用此方法，七种信号的识别率都可以达到86％以上，从而证明了该方法能有效地提高数字调制信号的识别性能。

关键词: 数字调制信号, 瞬时特征参数, 邻域信息, 粒子群算法（PSO）, 反向传播（BP）