一种基于自适应高斯神经网络的船舶噪声分类方法

提出了一种用于船舶噪声目标分类的自适应高斯神经网络分类方法.首先利用傅里叶变换对三类船舶噪声进行预处理,然后利用高斯函数特性,将其和神经网络结合构成自适应高斯神经网络对目标信号谱进行有效识别特征自动提取和分类.该方法获得的特征空间与以AR建模和子带平均功率诸方法获得的特征空间相比,类别之间的可分性好,类间聚集性强。分类结果令人满意,证明了该方法的优越性.

A classification method of noise radiated from ships based on adaptive Gauss neural networks

Based on adaptive Gauss neural networks,a classification method of noise radiated from ships is presented in this paper.Adaptive Gauss neural network is studied,and is used to extract automatically characteristics from noises radiated from three types of ships and to classify them after Fourier transform preprocessing.From the classified results,it is shown that the classified intervals are better and the number of the characteristics by using characterristics extracted by the adaptive Gauss neural network classifier is less than by using AR model and subband average power methods.The classified results are encouraging and this method is proved to be superior.