基于BP神经网络的波长扫描光纤干涉仪多目标信号处理技术

MULTI- OBJECTSDETECTION TECHNIQUESFOR WAVELENGTH SCANNING INTERFEROMETER BASED ON BP NEURAL NETWORK

本文提出了一种适用于波长扫描光纤干涉仪多目标绝对距离测量的方法 .这种方法将神经网络与数字滤波器等技术相结合 ,具有超过传统 FFT方法的目标分辨极限的分辨能力 ,可以工作在低的信噪比下 .这意味着可以降低对半导体激光器输出功率和扫描波长范围的要求 .计算机仿真实验验证了这种方法的性能 .

A signal processing method based on BP neural network is proposed for multi-objectabso- lute range detection of a wavelength scanning interferometry.Simulation result shows that,combined with digital filter,this method has a resolution higher than the limit of conventional FFT method. Hence,demand on both L D outputpower and wavelength-sweeping range can be reduced.