A Sewer Detection Method Based on Fiber-Optic Distributed Temperature Sensing and Wavelet Based Denoising

Abstract:

This paper proposed a method to determine background noise of fiber temperature sensing data based on wavelet denoising， and then detected in-sewer inflow events without disturbing sewer flow conveyance. This method proposed was validated using detected dynamic inflow events in an actual sewer system. It was found that different wavelet denoising algorithms provide background noises that span a wide range， and a noise threshold of about ±0.3 ℃ corresponds to the best identification of actual sewer and stormwater inflow events. Threshold rescaling is the dominant factor for algorithm employment， where the multi-level rescaling method is obviously superior to the non-rescaling and single-level rescaling method. Accordingly， optimized parameters for the wavelet denoising algorithm including wavelet function， threshold selection rules and threshold rescaling were proposed， to enhance the reliability of sewer detection.