In order to reduce the measurement error of Raman-based distributed temperature sensor (RDTS) caused by random noise, we have proposed an effective noise reduction method, which can reduce the time consumption of data processing due to the low time complexity of the method. We classified the signal as five models based on the waveform types, thus to filter the raw data according to the waveform types. Then, we established an experimental device to test the effects of different signal noise reduction methods, including maximum deviation (MD), root mean square error (RMSE), smoothness, and time consumption. Experimental results show that it can efficiently suppress random noise and reduce temperature measurement errors in RDTS compared with direct demodulation of the raw data, and significantly improve the curve smoothness compared with D-SVD and WT-Soft. The time consumption test based on three different devices shows that the time consumption of the proposed method is approximately 10% of the D-SVD method, and about 30% of the WT-Soft method. It proves that this method is an effective and fast noise reduction method in RDTS systems.

为了减少随机噪声对基于拉曼的分布式温度传感器（RDTS）的测量误差，我们提出了一种有效的降噪方法，由于该方法的时间复杂度低，可以减少数据处理的时间消耗。我们根据波形类型将信号分为五种模型，从而根据波形类型过滤原始数据。然后，我们建立了一个实验装置来测试不同信号降噪方法的效果，包括最大偏差（MD）、均方根误差（RMSE）、平滑度和时间消耗。实验结果表明，与直接解调原始数据相比，它可以有效抑制随机噪声并减少RDTS中的温度测量误差，并且与 D-SVD 和 WT-Soft 相比显着提高了曲线平滑度。基于三种不同设备的时间消耗测试表明，该方法的时间消耗约为 D-SVD 方法的 10%，约为 WT-Soft 方法的 30%。证明了该方法是RDTS系统中一种快速有效的降噪方法。