基于PSO-BP神经网络的DOA估计方法

提出利用粒子群算法优化BP神经网络来改善来波到达角估计性能的方法。传统的BP神经网络 易 陷入局部最优，因此采用粒子群算法对网络的权值和阈值进行优化，并将其应用到来波到达 角估计中。所提方法仅利用阵列协方差矩阵的第一行作为来波方位特征，与常用的 协方差矩阵 上三角特征相比，在不损失有效方位信息的基础上使特征维数极大降低。仿真实验 证明：同经典的RBF神经网络方法相比，基于所提方法的神经网络结构更简洁，泛化性能更 好，来波方位估计精度更高。

DOA estimation method based on PSO-BP neural network

Particle swarm optimization(PSO) is used for optimization of BP neural netw ork to improve the performance of direction of arrival(DOA) estimation. Due to t he fact that BP neural network is inclined to be trapped in local minimum poin t, a novel network-PSO based BP neural network is proposed and applied to DOA estimation. This method uses the first row of corre lation matrix instead of commonly used upper triangular half of the covariance m at rix, therefore the feature dimension is largely reduced without losing any DOA i nformation. Experimental results show that the performance of the proposed metho d is much better than that of classic RBF method in terms of neural netwo rk size, generalization and estimation precision.