Measurement and Recognition Algorithm for Distributed Fiber Sensor

In the measurement process, distributed fiber sensor is easily influenced by dynamic characteristics changes, cross sensitivity, abnormal signals and other uncontrollable factors to generate a larger deviation of r-squared of scattering spectra and affect signal analysis. Therefore, sensor hardly makes an effective recognition to intrusion events, thus divided into two perspectives. The dynamic compensation algorithm is put forward from reducing measurement errors and detects multi-road sensor information, and Cuckoo Newton Search (CNS) is combined with multi-channel data so as to analyze the signal data of scattering spectrum. On the other hand, from improving accuracy of recognition, a method based on bi-coherence spectrum for data decomposition and multi-domain vibration judgement and proportion judgement of characteristic quantity peak is came up to recognize intrusion events. Detailed experimental results showed that not only the multi-dynamic characteristics of sensor are improved but also the accuracy of information analysis. The observed accuracy for recognizing intrusion was over 85%.