中国激光, 2001, 28 (3): 232, 网络出版: 2006-08-10
基于神经网络的调频连续波光纤传感器扫描非线性复原技术
Nonlinear Recovering Technique Based on Neural Network for Frequency Modulated Continuous Wave Optical Fiber Sensors
摘要
分析了调频连续波光纤传感器中扫描非线性对距离测量的影响,并提出了一种复原技术。这种技术通过反向传播神经网络的学习来克服扫描非线性,可以在扫描源具有较强的非线性时获得对目标的精确估计。同时也研究了反向传播网络在线学习的问题,使这种方法可以适应环境的变化。
Abstract
Frequency-scanningnon-linearity influences range-detection accuracy of optical fiber sensors. In this paper,the influence is analyzed and a recovering technique is proposed. Non-linearity isrecovered through the learning process of backpropagation neural network. Simulationresults show that very accurate estimation is achieved even under severe non-linearity ofthe scanning source. The on-line learning of neural network is also investigated to makethis method more practical.
李阳, 冯正和, 龚建敏, 肖艳红, 廖延彪. 基于神经网络的调频连续波光纤传感器扫描非线性复原技术[J]. 中国激光, 2001, 28(3): 232. 李阳, 冯正和, 龚建敏, 肖艳红, 廖延彪. Nonlinear Recovering Technique Based on Neural Network for Frequency Modulated Continuous Wave Optical Fiber Sensors[J]. Chinese Journal of Lasers, 2001, 28(3): 232.