论文部分内容阅读
特征分析及提取是目标分类识别的重要环节.首先应用非线性分析方法分析了舰船噪声的极限环现象,结果表明振动噪声作为舰船噪声的主要成份,其极限环在相空间上存在倍周期或混沌行为.其次,利用分形维数和分布密度比来描绘舰船噪声在相平面上极限环的奇异性和空间形状,并给出了一种新的分维数计算算法.最后,以此作为舰船目标的特征参数送入神经网络分类器用于分类识别水面和水下两大类目标.实验结果表明。从噪声极限环中提取的非线性特征能较准确地区分我们现有的水面和水下两大类目标.、为舰船噪声信号的特征提取提供了一条新的思路
Feature analysis and extraction is an important part of target classification. Firstly, the limit cycle phenomenon of ship noise is analyzed by nonlinear analysis method. The results show that vibration noise is the main component of ship noise, and its limit cycle has double cycle or chaos in phase space. Secondly, using the fractal dimension and distribution density ratio to describe the singularity and space shape of the limit cycle of the ship noise on the phase plane, a new algorithm of fractal dimension calculation is given. Finally, it is used as the characteristic parameter of the ship target into the neural network classifier for classification and identification of two major categories of water surface and underwater target. Experimental results show. The non-linear features extracted from the noise limit ring can more accurately distinguish between our existing surface and underwater targets. , Which provides a new idea for the feature extraction of ship noise signal