论文部分内容阅读
为了提高矢量水听器阵列对窄带信号的DOA估计精度,运用果蝇算法优化广义回归神经网络,通过对阵列协方差矩阵实值化,并提取信号子空间的基作为样本特征进行网络训练,构建了果蝇算法优化下的广义回归神经网络,实现了基于矢量水听器阵列的水下声源的DOA估计.仿真实验结果表明,方法泛化性能较好,能解决输入维数过大的问题,且运行时间短,DOA估计精度高,具有较强的工程应用价值.
In order to improve the accuracy of DOA estimation of narrow-band signals by vector hydrophone array, the generalized regression neural network (RBFNN) is optimized by using Drosophila algorithm. The array covariance matrix is realized and the signal subspace base is extracted as the sample feature for network training. A generalized regression neural network optimized by Drosophila algorithm is used to realize the DOA estimation of underwater sound sources based on vector hydrophone array.The simulation results show that the method has good generalization performance and can solve the problem of excessive input dimension , And the running time is short, DOA estimation accuracy is high, and has strong engineering application value.