摘要
研究了船舶的辐射噪声的分形特征和噪声谱 AR模型特征 ,并利用这两个特征作为区分船舶的依据 ;通过聚类分析 ,发现本文提取的特征具有良好的可分性。文中还详尽研究了 Max- Min模糊神经网络 ,讨论了各个参量对隶属度函数形成的作用 ,并提出了几项改进措施 ,解决了原有方法的固有问题 ,通过仿真与实船数据验证分类效果有明显的提高。
The radiated noise of a ship is analyzed, the Maples′ method is applied to modeling the simplified spectrum of the ship noise. This paper studies the relationship between the fractal characteristics and the size of the ship. Based on these work, it takes advantage of these characteristic to distinguish the types of the ships. At the same time, a fuzzy neural network is used as a classifier, an improved project is put forward, and particular superiorities are discussed. Finally, satisfactory results are gained in an experiment.
出处
《数据采集与处理》
EI
CSCD
2000年第2期142-147,共6页
Journal of Data Acquisition and Processing
基金
"九五"国防科工委重点预研资助
关键词
船舶噪声
谱分析
AR模型
特征描述
模糊分类
ship noise
spectrum analysis
fractal dimension
fuzzy neural network
autoregressive model
