一种鲁棒的被动声目标识别方法

A Robust Method in Passive Acoustic Target Recognition

针对噪声环境下一些声目标识别技术性能严重下降的问题，提出了一种基于一阶补偿的径向基函数网络模型，并给出了该网络参数选择方法。实际应用表明，这种神经网络对噪声具有较强的鲁棒性，与传统的径向基函数网络相比，其识别性能等同于信噪比提高大约１０～１５ｄＢ。

A more robust passive acoustic target recognizer is researched to solve the problem of performance degradation in a noisy environment. The model of RBFN with first\|order equalization is presented and the algorithms for estimating the model parameters are given. Experimental results show that this model is more robust to noise, with respect to a conventional RBFN recognizer, this network makes an improvement in recognition performance which is equivalent to about 15dB gain in SNR.