基于反向传播神经网络的海杂波参数估计

研究分析海杂波、评估弹载导引头在不同海况下的打击精度具有重大意义。针对传统统计方法的海杂波参数估计易存在脱离实况海杂波物理特征的问题，提出基于反向传播（BP）神经网络的参数估计法。利用海杂波幅度分布特性和时间相关性，建立基于K分布的时间与空间相关海杂波模型；重点分析形状参数、尺度参数、杂波速度均方根、平均多普勒频移4个模型参数对海杂波混沌特性、分形特性的影响，总结出模型参数与物理特征之间的定性关系；利用BP神经网络充分挖掘参数与物理特征间的定量关系，并对混沌特性、分形特性进行预测，决定系数为0.985、0.952. 以实测海杂波数据为例，比较BP神经网络、最大似然估计和矩估计法的模型参数，验证了该方法可以较好地贴近真实海杂波的物理特征。

Sea Clutter Parameter Estimation Based on BP Neural Network

Sea clutter is studied and analyzed for evaluating the accuracy of missile-borne synthetic aperture radar under different sea conditions. A parameter estimation method based on BP neural network is proposed to solve the problem that the sea clutter parameter estimation based on traditional statistical methods is prone to be divorced from the actual sea clutter. The amplitude distribution characteristics and temporal correlation of sea clutter are used to establish a K-distribution-based sea clutter model. The influences of four model parameters, such as shape, scale, mean square root of clutter speed, and mean Doppler shift, on sea clutter chaos and fractal characteristics are analyzed, and the qualitative relationship among model parameters and physical characteristics is summarized. On this basis, BP neural network is used to fully explore the quantitative relationship among parameters and physical characteristics, and predict the chaos characteristics and fractal characteristics. The determinate coefficients are 0.985 and 0.952. The model parameters of BP neural network, maximum likelihood estimation and moment estimation method are compared by taking the measured sea clutter data as an example. The results show that the proposed model can be well close to the physical characteristics of actual sea clutter. Key