基于神经网络二维PSD线性优化方法研究
|
摘要
位置敏感器件(PSD)具有性价比高、外围电路简单等优点,在角度等高精度测量领域具有广泛应用.但PSD的非线性限制了其精度,针对PSD(位置敏感器)固有的非线性,分析了二维PSD非线性特征及对测量结果的影响,提出了一种基于神经网络的二维PSD线性优化的算法,使用两个稳含层对神经网络进行训练,建立非线性映射关系,对二维PSD实现非线性补偿,通过MATLAB仿真实验对比,结果显示该算法能有效地减少二维PSD非线性的影响,且精度较高,收敛速度快,提高了测量数据的准确性.
Position-sensing devices( PSD) have the advantages of cost-effectiveness,simple peripheral circuits,and are widely used in high-precision measurement fields such as angles. However,the nonlinearity of the PSD limits its accuracy. Based on the inherent nonlinearity of the PSD( Position Sensor),the nonlinear characteristics of the two-dimensional PSD and the influence on the measurement result are analyzed. A two-dimensional PSD linear optimization based on a neural network is proposed. The algorithm uses two stable layers to train the neural network,establishes a nonlinear mapping relationship, and achieves nonlinear compensation for the two-dimensional PSD. The results of the MATLAB simulation show that the algorithm can effectively reduce the nonlinearity of the two-dimensional PSD. Impact,high accuracy,fast convergence,improve the accuracy of measurement data.
Position-sensing devices( PSD) have the advantages of cost-effectiveness,simple peripheral circuits,and are widely used in high-precision measurement fields such as angles. However,the nonlinearity of the PSD limits its accuracy. Based on the inherent nonlinearity of the PSD( Position Sensor),the nonlinear characteristics of the two-dimensional PSD and the influence on the measurement result are analyzed. A two-dimensional PSD linear optimization based on a neural network is proposed. The algorithm uses two stable layers to train the neural network,establishes a nonlinear mapping relationship, and achieves nonlinear compensation for the two-dimensional PSD. The results of the MATLAB simulation show that the algorithm can effectively reduce the nonlinearity of the two-dimensional PSD. Impact,high accuracy,fast convergence,improve the accuracy of measurement data.
引文
[1]李博,高艺,王红平,等.高精度PSD线性方法与实验研究[J].长春理工大学学报,2013,36(1):36-39.
[2]杨孝敬.二维PSD非线性误差校正算法研究[D].郑州:河南工业大学.
[3]宋庆.基于2D-PSD的长导轨直线度测量系统的研究[D].长春:长春理工大学.
[4]冯华君.主动式PSD自动对角光学系统研究[J].光学仪器,2010,20(1):13-17
[5]姚明海.改进的遗传算法在优化BP网络权值中的应用[J].计算机工程与应用,2013,49(24):49-54.
[6]史春玉,李田泽,仝其丰,等.基于改进共轭梯度算法的二维PSD非线性修正[J].山东理工大学学报,2017,31(2):27-30.
[7]王晓东,吴威,兰玉杰,等.光电位置敏感元器件及其信号的检测[J].电测与仪表,2008,12(1):24-26.
[8]张欣婷,亢磊,吴倩倩,等.基于BP优化算法的PSD非线性校正[J].应用光学,2016,37(3):415-418.
[9]刘春,马颖.遗传算法和神经网络结合的PSD非线性修正[J].电子测量与仪器学报,2015,29(8):1157-1163.
[10]李倩.基于遗传BP神经网络的二维PSD非线性研[D].淄博:山东理工大学.
[11]姚建国,阎文,梁光建,等.基于一维PSD误差分段的BP网络非线性校正[J].传感器与微系统,2008,28(1):24-26.
[2]杨孝敬.二维PSD非线性误差校正算法研究[D].郑州:河南工业大学.
[3]宋庆.基于2D-PSD的长导轨直线度测量系统的研究[D].长春:长春理工大学.
[4]冯华君.主动式PSD自动对角光学系统研究[J].光学仪器,2010,20(1):13-17
[5]姚明海.改进的遗传算法在优化BP网络权值中的应用[J].计算机工程与应用,2013,49(24):49-54.
[6]史春玉,李田泽,仝其丰,等.基于改进共轭梯度算法的二维PSD非线性修正[J].山东理工大学学报,2017,31(2):27-30.
[7]王晓东,吴威,兰玉杰,等.光电位置敏感元器件及其信号的检测[J].电测与仪表,2008,12(1):24-26.
[8]张欣婷,亢磊,吴倩倩,等.基于BP优化算法的PSD非线性校正[J].应用光学,2016,37(3):415-418.
[9]刘春,马颖.遗传算法和神经网络结合的PSD非线性修正[J].电子测量与仪器学报,2015,29(8):1157-1163.
[10]李倩.基于遗传BP神经网络的二维PSD非线性研[D].淄博:山东理工大学.
[11]姚建国,阎文,梁光建,等.基于一维PSD误差分段的BP网络非线性校正[J].传感器与微系统,2008,28(1):24-26.