Research on multi-fiber speckle strain sensors based on structure-data fusion
Lists
Only you see this
Abstract
Abstract The demand for high performance strain sensors with both high accuracy and efficient computation is escalating in fields such as aerospace, civil engineering, and machinery manufacturing. However, integrating high measurement precision and low computational complexity in a single sensor remains a challenge. Here, we propose a multi-fiber speckle strain sensor based on structure-data fusion. The sensor is innovatively designed by optimizing the spatial topological structure of multiple fibers, with a single-turn coaxial fiber bundle as the core sensing architecture, consisting of one central transmitting fiber (TF) and six surrounding receiving fibers (RFs). When coherent light from the TF is reflected by the measured object, the RFs capture speckle patterns with strong correlation. Through coding preprocessing of these speckles and analysis via a backpropagation neural network system (including six Sub-BPNNs and one Gross-BPNN), strain measurement is achieved. As the sensor detects strain changes, the structured speckle field formed by the multi-fiber layout encodes abundant strain-feature information, reducing the complexity of subsequent signal processing. The fusion of spatial structure information and speckle data features effectively mitigates algorithmic uncertainty. Benefiting from this structure-data fusion strategy, the sensor achieves high measurement accuracy (MAE = 0.1133, RMSE = 0.1591) with low computational requirements. The proposed multi-fiber speckle sensing methodology shows significant potential for applications in complex engineering environments.
BibTeX
@inproceedings{yin2025research,
author = {Ye Yin and Rui Meng and Binghui Jia and Zhuo Jin},
title = {{Research on multi-fiber speckle strain sensors based on structure-data fusion}},
booktitle = {Measurement},
year = {2025},
doi = {10.1088/1361-6501/ae1bc4},
}Loading related content...