本文提出一种由广义小波神经网络实现船用雷达跟踪中航迹外推的自适应新方法 .由S(sigmoid)函数构造的尺度函数和小波作为网络中神经元的激励函数 ,隐层节点数由小波分解次数和处理信号维数决定 ,输出层采用局部连接方式以解决多维信号的不利影响 .理论证明 ,广义小波神经网络的鲁棒性在一定条件下优于BP网络 .仿真表明 ,该方法的在线处理运算量不随所跟踪的运动目标模型的复杂性而增加 ,并且对变加速和急转弯运动目标具有较高的跟踪精度 . In this paper, an adaptive new method of generalized wavelet neural network to track extrapolation in marine radar tracking is proposed.The scale function constructed by S (sigmoid) function and wavelet as the excitation function of neurons in the network, the number of hidden layer nodes is determined by wavelet The number of decompositions and the dimension of signal to be processed are determined, and the output layer adopts the local connection method to solve the adverse effect of multi-dimensional signal.It is proved theoretically that the robustness of the generalized wavelet neural network is better than that of the BP network under certain conditions. Simulation shows that the online The amount of processing operations does not increase with the complexity of the moving target model being tracked and has a high tracking accuracy for variable speed and sharp turning sports targets.