为了解决由于位置敏感探测器有效光敏面过小导致的激光通信设备中信标光探测视场过小的问题,提出了一种基于BP神经网络的校正方法,并确定了一种适合FPGA实现的BP神经网络结构。该方法利用神经网络良好的非线性映射能力逼近非线性函数,从而实现非线性校正,增加光敏面可用面积达到增加探测视场的效果。仿真结果表明2-15-15-2结构且传递函数选择Log-sigmod+Log-sigmod函数的BP神经网络可有效抑制位置敏感探测器非线性的影响且易于硬件实现。

In order to solve the problem that the field of view of the beacon optical detection in the laser communication device is too narrow due to the small effective photosensitive surface of the position-sensitive detector, a correction method based on BP neural network is proposed, and a BP neural network structure which is suitable for FPGA implementation is determined. This method uses the good nonlinear mapping ability of neural network to approximate the nonlinear function so as to achieve the non-linear correction and increase the available area of the photosensitive surface to increase the effect of detecting the field of view. The simulation results show that the BP neural network with 2-15-15-2 structure and the function of Log-sigmod + Log-sigmod transfer function can effectively restrain the non-linearity of the position-sensitive detector and can be easily implemented in hardware.