论文部分内容阅读
针对目前相位敏感光时域反射计(?-OTDR)分布式光纤扰动传感系统难以对扰动事件进行有效判别的问题,提出一种基于多特征参量的扰动模式识别方法。作为识别依据的多特征参量包括平方差、短时过电平率、短时傅里叶变换以及扰动持续时间,对以上特征参量利用加权平均型综合评判函数进行模式识别,可以有效识别出应力破坏(敲击)、攀爬、浇水(模拟下雨环境)以及轻度碾压(模拟非破坏性人为扰动)四种扰动模式。对60组共240个实验样本进行识别,对以上四种扰动事件的识别率分别达到91.2%,88.3%,90%和86.7%。该方法克服了现有单一特征参量识别模式受限的问题,提高了扰动事件识别的准确性。
Aiming at the problem that the phase-sensitive OTDR (? -OTDR) distributed optical fiber disturbance sensing system is hard to discriminate perturbation effectively, a disturbance pattern recognition method based on multi-characteristic parameters is proposed. The multi-characteristic parameters used for identification include square error, short-time over-level rate, short-time Fourier transform and duration of disturbance. Pattern recognition of the above characteristic parameters using the weighted average comprehensive evaluation function can effectively identify the stress failure (Percussion), climbing, watering (simulated rainy environment), and light rolling (simulating nondestructive man-made disturbance). A total of 240 experimental samples of 60 groups were identified, and the recognition rates of the above four disturbance events were 91.2%, 88.3%, 90% and 86.7% respectively. The method overcomes the problem of limited identification mode of the existing single feature parameters and improves the accuracy of identification of disturbance events.