改进型BP神经网络的2维PSD非线性校正

2-D nonlinear corretion in an improved BP neural network

为了减少位置敏感传感器(PSD)的非线性的影响,分析了PSD的工作原理及其非线性成因,提出一种基于Levenberg-Morquardt算法改进的反向传播(BP)神经网络方法进行非线性修正,并进行了理论分析和MAT-LAB仿真比较。结果表明,改进的BP神经网络方法能有效地减少非线性影响,且相对传统的BP神经网络而言,收敛速度更快,使修正后的PSD器件在非线性区里获得与线性区近似的线性度。这一结果对PSD更好的应用是有帮助的。

In order to reduce the effect of nonlinearity of a position sensitive detector（PSD）, after analyzing its working principle and the reasons of nonlinearity formation, nonlinearity correction was carried out in an improved back propagation（BP） neural network based on Levenberg-Morquardt algorithm. MATLAB simulation results show that the improved BP neural network can reduce nonlinearity more effectively, and converge faster than a traditional BP neural network. After revision the PSD obtains approximate linearity in non-linear area within the linear area. This result is helpful for PSD application.