基于高阶累积量的水声噪声检测与识别

DETECTION AND RECOGNITION OF SHIP NOISE BASED ON HIGHER-ORDER CUMULANT

本文基于高阶统计量理论研究了海洋环境噪声和船舶辐射噪声的高斯性 ,同时基于高阶累积量T2 研究了非高斯背景下非高斯信号的一种检测方法 ,并用于海洋环境噪声下检测船舶辐射噪声的方法研究中。另外研究了三类船舶辐射噪声双谱中对角切片和非对角切片特征 ,并将其作为表征船舶辐射噪声的特征向量 ,经过一定的实船实验得到了较为满意的结果。

The gaussianity of water acoustic signal was studied by higher order statistics. Some rules were found. They show that ship noise can be regarded as non Gaussian signals. Calm sea noise late at night is Gaussian and sea noise in the daytime must be deemed to be non Gaussian. Based on these characters, the method of detecting the non Gaussian signal in a non Gaussian background is put forward by using the higher order cumulant T 2 , and ulteriorly making use of this method to solve the problem detecting the ship noise from the sea noise. In addition, based on the higher order cumulant of the ship noise, the characters of diagonal slices and non diagonal slices are abortively studied and on second thoughts they are regarded as character eigenvectors for classifying three types of the ship noise. By using the IMMNN classifier, fine results have been obtained.