水下混沌背景中的瞬态声信号检测法研究

水下瞬态声信号中蕴含着目标的特征信息，但其突发性强、持续时间短致使检测难度很大。为解决瞬态信号检测的问题，提出了混沌背景中瞬态冲击信号的RBF神经网络检测法。建立了混沌背景噪声的一步预测模型，通过预测误差的变化来检测瞬态信号。分别以Lorenz系统和Logistic系统作为混沌背景噪声进行了仿真，证明检测方法的有效性，并在Lorenz系统背景检测中加入白噪声来检验该方法抗白噪声干扰的能力，结果表明该方法对白噪声敏感；在理论研究的基础上通过对外场试验数据的处理验证了该方法的有效性，并在实际测量数据中加入混沌背景噪声，通过改变信噪比检验了该方法在不同信噪比情况下的性能。

The Detection of Underwater Transient Acoustic Signal Under Chaotic Background

There were some important information about the target carried by the transient acoustic signal. But the transient acoustic signal underwater often breaks out abruptly and only last for a few periods, so it is difficult to be detected. In order to detect the underwater transient acoustic signal, the RBF neural network detection method under the chaotic background was put forward. Due to the characters of RBF network and chaotic system, a one-step predicting model was set up. The simulation experiment was carried out based on both Lorenz system and Logistic system, and white noise was also put into the chaos background to validate the detecting ability of this method. The method was proved to be sensitive to the white noise. Based on the research of the theory, several data gained from the experiment on the lake were processed to prove the validity of the method presented above. The data measured in the experiment was add to a sect of stronger chaos background noise to research the performance of the detection method under different SNR