一种基于神经网络的激光制导炸弹目标环境补偿方法

Target Environment Compensation Method Based on Neural Network for Laser Guided Bomb

根据激光制导炸弹的工作原理,分析了环境光对四象限光电探测器的影响,得出了存在环境光时输出信号是非线性的。利用神经网络所具有地能够以任意精度逼近非线性函数的能力,采用神经网络共轭梯度法对四象限光电探测器的输出信号进行了非线性修正,实现四象限探测器的输出与实际光点位置之间的映射,在神经网络的输出端得到线性响应。并且通过仿真分析证明了该方法是有效的。结果表明,修正后的四象限探测器能在较大的位置范围内输出高线性度的信号,从而提高激光制导炸弹制导精度和稳定性。

According to the principle of the laser guided bomb and analyzing the influence of environment light, and it can be included that the output signal of the four-quadrant detector is non-linear if environmental light exits. The neural network can approach the nonlinear function by the free precision, so the neural network conjugate gradient method is adopted to achieve the non-linear revision of four-quadrant detector output signal. Mapping between four-quadrant detector output and the actual luminous spot position is achieved. The linear response is obtained in the neural network outport. Simulation analysis proves this method is effective. The result shows that the modified four-quadrant detector can output linear signal in the wide range,and the accuracy and stability of the laser-guided bomb are greatly improved.