【摘要】 根据光电位置敏感器件(Position sensitive detector,PSD)的原理和光点位置方程分析了其非线性成因和特点,提出了用神经网络的共轭梯度算法对PSD的非线性进行补偿。以网络总体平均误差为目标函数,以权值和阈值为设计变量,采用真实的梯度法和共轭梯度法对网络参数优化,实现了权值和阈值的快速准确计算,并将该方法应用于PSD线性化,从而使PSD的B区(边缘区域)获得了与A区(中央区域)近似的线性度。故在不增加成本,不改变测量设备复杂度的情况下,扩大了测量范围,提高了B区的测量准确度及数据的置信度。更多还原

【Abstract】 The principle of position sensitive detector(PSD) is briefly introduced. The causes of non-linearity of PSD are analyzed. Through defining network average error as objective function, weights and thresholds as design variable, network parameters are computed quickly by gradient and conjugate gradient optimizations. Conjugate gradient optimum algorithm for neural network to correct the non-linearity of PSD after pre-calibration are provided. This method is applied to linearity of PSD. The correction leads to prominent improvement of linearity of B-area, and the usable area of PSD is thus extended. The reliability of data is also improved without increasing the complexity of hardware.更多还原