Research on multi-source vibration identification method of distributed fiber sensing based on improved fast independent component analysis

Distributed acoustic sensing (DAS) is a technology that uses optical fiber as a sensing unit to detect external vibration signals. Due to the high resolution and high sensitivity of DAS, it has great application potential in the detection of vibration events. However, high detection performance will bring limitations to DAS in multi-source detection. In order to solve the problem of poor multi-source and multi-target recognition ability of DAS in complex multi-source environments, a multi-target recognition processing method of DAS system is proposed. Firstly, the acquired signal is pretreated to suppress the influence of system noise. Secondly, based on the independence of different events, a multi-source vibration location identification method based on the principal eigenvalue analysis of the observation matrix is proposed to realize the identification of the number and location of the multi-source vibration. Finally, the dimension of the acquired signal is reduced according to the number of sound sources, and the multi-source vibration signal recognition method based on amplitude uncertainty suppression and fast independent component analysis is proposed to realize the multi-source vibration signal recognition. The experimental results show that the proposed method has achieved good processing effect in both simulation and actual measurement, and the relative error is kept below 0.3. In comparison experiments, the proposed method is superior to principal component analysis, EMD and EWT in location recognition and signal recognition effect. This method provides technical support for DAS application in complex multi-source environment and improves the practical application value of DAS.