US 12,345,567 B2
Road surface conditions detection by distributed optic fiber system
Yuheng Chen, South Brunswick, NJ (US); Ming-Fang Huang, Princeton, NJ (US); Ting Wang, West Windsor, NJ (US); and Jingnan Zhao, Edison, NJ (US)
Assigned to NEC Corporation, Tokyo (JP)
Filed by NEC Laboratories America, Inc., Princeton, NJ (US)
Filed on Nov. 15, 2022, as Appl. No. 17/987,007.
Claims priority of provisional application 63/280,258, filed on Nov. 17, 2021.
Prior Publication US 2023/0152150 A1, May 18, 2023
Int. Cl. G01H 9/00 (2006.01); G01P 3/36 (2006.01); G06N 3/08 (2023.01); G06N 20/10 (2019.01)
CPC G01H 9/004 (2013.01) [G01P 3/36 (2013.01); G06N 3/08 (2013.01); G06N 20/10 (2019.01)] 4 Claims
OG exemplary drawing
 
1. A method for road surface conditions detection using distributed fiber optic sensing (DFOS), said method comprising:
providing a length of optical sensor fiber, said length of optical sensor fiber being positioned parallel to the road surface;
providing a DFOS interrogator in optical communication with the optical sensor fiber, said DFOS interrogator configured to generate optical pulses, introduce the generated pulses into the length of optical sensor fiber, and receive backscattered signals from the length of the optical sensor fiber; and
providing an intelligent analyzer configured to analyze the received backscattered signals and determine from the backscattered signals, vibrational activity occurring at locations along the length of the optical sensor fiber;
determine vehicle speed of vehicles operating on the road surface from the backscattered signals using Hough transform methods;
determine locations of potholes or cracks in the road surface; and
outputting an indicium of the pothole or crack locations;
wherein the intelligent analyzer determines the vibrational activity occurring at locations along the length of the optical sensor fiber by generating annotated waterfall plot images from the backscattered signals and extracting features of the annotated images using local binary pattern histograms, and by performing a principal component analysis (PCA) on the annotated images such that a dimension is reduced and employing a support vector machine methodology as a classifier to generate a trained model; and
wherein a convolutional neural network (CNN) is used to train binary classification data to generate a trained model.