论文部分内容阅读
提出了一种基于激光拍频测量实现光纤光栅温度传感解调的方法,并构建了三层BP神经网络模型对温度传感数据进行优化。该方法分别采用线性啁啾光栅(CFBG)和传感光纤光栅(FBG)作为光纤激光系统的反馈腔镜,测量激光器拍频随传感光栅温度的变化实现温度传感。为减小CFBG非线性时延及时延抖动引起的温度测量误差,采用三层BP神经网络模型,将所测的拍频频率与对应温度经过BP神经网络训练。实验中,重复测量10次,得到10组拍频频率/温度数据。随机选择9组频率数据作为训练校正集,送入三层BP网络模型的输入层作为网络输入值,其相应的实际温度值作为网络的输出值,训练网络的权值和阈值,直至满足设定的目标使网络的参数及结构最优化。另一组作为验证样本集,测试该网络模型的实用性。此组数据的温度灵敏度和相关系数分别为37.89kHz/℃和99.767%,对该组数据训练温度校正及预测,其相关系数达到99.95%。结果表明,利用三层BP神经网络算法对实验数据进行校正,能够有效地提高系统的测量精度。
A method of temperature sensing demodulation based on laser beat measurement is proposed, and a three-layer BP neural network model is constructed to optimize the temperature sensing data. The method uses linear chirped grating (CFBG) and sensing fiber grating (FBG) respectively as the feedback cavity mirror of the fiber laser system, and measures the laser beat frequency to achieve the temperature sensing with the change of the sensing grating temperature. In order to reduce the error of temperature measurement caused by nonlinear delay and jitter of CFBG, a three-layer BP neural network model is adopted to train the measured beat frequency and corresponding temperature through BP neural network. In the experiment, the measurement was repeated 10 times to obtain 10 sets of beat frequency / temperature data. Nine groups of frequency data are randomly selected as the training correction set and input into the input layer of the three-layer BP network model as the input value of the network. The corresponding actual temperature value is used as the output value of the network to train the weights and thresholds of the network until the setting The goal is to optimize the parameters and structure of the network. Another group as a verification sample set to test the practicality of the network model. The temperature sensitivity and correlation coefficient of this set of data were 37.89kHz / ° C and 99.767%, respectively. Correction coefficients and their correlation coefficients were 99.95%. The results show that using the three-layer BP neural network algorithm to correct the experimental data can effectively improve the measurement accuracy of the system.