光纤传感器的神经网络光强补偿及非线性校正

摘要

This paper established and constructed a two-input single-output BP neural network to realize the light intensity compensation and nonlinear correction of optical fiber displacement sensor. It discussed the advantages of BP network based on L-M optimization algorithm, indicated the weight modified method and neural network training steps, clarified the principle of the sen-sor' s light intensity fluctuation compensation and non-linear correction of using the neural network. Finally, it used Matlab software on computer to program the neural network, and applicant the network in the actual measurement. The results show that the network can achieve a good compensation for light intensity and nonlinear correction of sensors.%为实现光纤位移传感器的光强补偿及非线性校正,确立并构造了两输入单输出的BP神经网络.讨论了基于L-M优化算法的BP网络的优点,说明了神经网络的权值修正方法及网络的具体训练步骤,阐明了用神经网络实现光纤位移传感器光强波动补偿及传感器非线性校正的原理.最后使用Matlab实现了神经网络,并将该网络应用于实际测量,结果表明该网络很好地实现了传感器的光强补偿及非线性校正.