Dynamic strain measurement in Brillouin optical correlation-domain sensing facilitated by dimensionality reduction and support vector machine
Yuguo Yao, Yosuke Mizuno
Abstract
Brillouin optical correlation-domain sensing enables high-speed Brillouin gain spectrum (BGS) measurement at random positions along the optical fiber. To extract the Brillouin frequency shift (BFS) that reflects the real-time strain information, machine learning methods of principal components analysis (PCA) and support vector machine (SVM) are used in the signal processing for the BGSs. The performances of dimensionality reduction by PCA and SVM based on classification and regression are analyzed and compared. The experiment demonstrates an 8 kHz BGS acquisition repetition rate and an average BFS extraction time of 0.0104 ms, which is 27.3 times faster than the conventional method with no PCA. The proposed methods realize a real-time dynamic strain measurement at the frequency of 40 Hz.