海杂波背景下基于神经网络的目标检测

Targets Detection Based on Neural Network under Background of Sea Clutter

为了检测海杂波背景下的目标回波信号，提出了一种基于神经网络的目标检测方案。设计径向基函数（RBF）网络对海杂波的混沌动力学特征进行学习和预测，并对预测误差作功率谱分析，可以检测出目标信号并提取其速度和距离信息。应用实际参数在岸基和舰载雷达条件下的仿真结果表明该方法优于感知器门限分类检测和目标信息的二维FFT提取，检测相对误差控制在10^-3～10^-4量级。仿真进一步给出了两种条件下的检测最低信噪比。

To detect echo wave from targets under the background of sea clutter, a new method based on neural network was brought up. Radial based function network can be trained to predict the chaotic dynamic characters of sea clutter. By analyzing the power spectrum of predicting error, the target detection could be made and its speed and distant information. were extracted The emulation results on the conditions of shore-based and ship-borne radar with practical parameters prove to be better than perceptron threshold detection and two-dimension FFT for targets＇ information. The relative error of detection is controlled within 10^-3～10^-4 magnitude. Then the lowest signal-to-noise ratios on each condition were given.