基于EMD与三阶累积量的水声瞬态信号检测

摘要

提出了基于经验模态分解(EMD)与三阶累积量的水声瞬态信号检测方法.首先根据EMD方法的滤波器特性将待检信号在频域内分成一系列的本征模态函数(IMF)分量,并根据能量法选择信号占主导地位的IMF；然后运用高阶累积量抑制高斯信号的特性,计算IMF分量三阶累积量对角切片的短时估计,并构造检测函数,对检测函数时行包络检波,作为检测标准；最后用仿真数据对该方法进行了验证,结果表明能在较低信噪比下检测出目标信号的出现时刻和大致频率,具有一定的工程应用价值.%A detection method of underwater acoustic transient signals based on EMD and third order cumulant was presented in the paper. First, EMD method was employed to decompose the signal into a group of IMFs according to their frequencies. The IMF with bigger energy power was chosen as the useful component. Then third-order cumulant was used to depress the white Gaussian noise and the short-time estimation of third-order cumulant diagonal slice can be received during the process. A detection function which is based on short-time estimation and its envelope were recognized as the detection statistic to determine whether there are transient signals. At last, simulated data were used to test the effect of the method. The results illustrate that the detection method based on EMD and third-order cumulant can detect the underwater acoustic transient signals at lower SNRs, which is very valuable in the engineering application.