论文部分内容阅读
本文系统深入地分析研究了舰船噪声信号的时域波形结构特征，利用舰船噪声信号的过零点、峰间幅值、波长差、波列面积分布以及时域特性提取技术，将原始舰船噪声信号时域波形分类信息表达成了11维的分类特征向量，同时设计了结构自适应模型聚类神经网络分类器，对提取的舰船噪声分类特征向量进行分类．训练样本集平均识别率达96．72％；测试样本集平均识别率达88．39％，分类实验结果令人满意
In this paper, the time-domain waveform structure characteristics of the ship noise signal are analyzed and studied in depth. Based on the zero crossing, the peak amplitude, the wavelength difference, the wavefront area distribution and the extraction of the time domain characteristics of the ship noise signal, The time-domain waveform classification information of the noise signal is expressed as an 11-dimensional classification feature vector. At the same time, a structure-adaptive model clustering neural network classifier is designed to classify the extracted ship noise classification feature vectors. The average recognition rate of training samples was 96.72%, the average recognition rate of test samples was 88.39%, and the classification experiment was satisfactory