Abstract: Brillouin optical time domain reflectometer/analysis (BOTDR/A) is one kind of important distributed optical fiber sensing technology. Using the BOTDR/A-based distributed optical fiber technology to detect deformation of pile foundation, the vast amounts of detection data need smoothing and denoising. The paper introduces most frequently used embedded techniques of the sensing optical fiber, the characteristics of detection data, the role of massive data smoothing and denoising on the detection data handle and analysis. We developed the concept and proposed the implementation process of the feature point compression algorithm (a new smoothing and denoising method). Based on data measured from practical engineering, the smoothing and denoising effect of the method is analyzed, and its applicability is also verified. The results show that the proposed method is simple and efficient in processing the data from pile foundation detection, compared with the conventional algorithm, it can smooth and denoise data without losing the characteristics trend of detected data, and can achieve satisfactory results as expected. It can be applied to the distributed data processing of optical fiber pile foundation detection.

Key words: pile foundation, distributed optical fiber detection, feature point compression algorithm, data smoothing and denoising