基于深度学习测量水温的多模光纤传感器

打开文本图片集

中图分类号：O436 文献标志码：A 文章编号： 1000-5013（2026）01-0076-07

Abstract：By utilizing the characteristic thatthe speckle image generated bya laser beam transmiting through a multimode optical fiber vary with temperature，the multimode fiber is immersed in water to enable water temperature monitoring. A convolutional neural network（CNN） model is designed， a speckle image dataset corresponding to water temperatures ranging from 1. 6 to 60.0°C is constructed to train the system for monitoring the temperature water level in multimode fiber. The results show that within a deviation range ±1 ，the proposed method provides an average temperature monitoring accuracy 98.66% ，which fers a high-accuracy，wide-range solution for underwater temperature sensing monitoring.

words：speckle image；convolutional neural network；high precision temperature monitoring；multimode fiber

因体积小、灵敏度高、抗电磁干扰等优点，光纤传感器在温度传感领域[1-2]得到广泛应用，尤其在水下环境监测[3]、海洋资源开发[4]、水下工程[5]等领域，长距离的光纤传感器可以实现大范围温度监测，具有重要的应用价值[6]。（剩余6697字）