Error correction of temperature measurement data obtained from an embedded bifilar optical fiber network in concrete dams

Abstract Conventional distributed optical fiber systems employed for internal temperature sensing in concrete dams suffer from low survival rate, insufficient measurement data, and low-accuracy temperature measurements that are subject to large fluctuations. Therefore, the present study proposes a bifilar optical fiber embedding technique running within each of the concrete dam blocks forming a dam. The proposed fiber optic network design is based on multiple considerations, including the embedding requirements of optical fiber to ensure good system longevity, the requirements of internal temperature monitoring in concrete dams to ensure the capture of adequate monitoring data, and the temperature measurement principle of distributed temperature sensing (DTS) systems to develop effective error correction methods. In terms of error correction, the light intensity attenuation caused by transmission- and wavelength-related losses is calculated based on the symmetrical temperature measurement points of the primary and secondary strands of the embedded bifilar optical fiber in conjunction with error correction theory to correct for temperature measurement fluctuations and thereby ensure the capture of accurate and stable temperature measurement data. Experimental error correction test results demonstrate that the mean absolute error (MAE) and root mean squared error (RMSE) of the error corrected temperature data obtained under various temperature conditions are reduced by 0.28 °C and 0.34 °C, with fluctuations that are decreased by 53.4% and 52.4%, respectively. The results derived from an actual engineering application under construction demonstrate that the survival rate of the embedded bifilar optical fibers is 100%. In addition, the average MAE and RMSE values of the temperature data obtained by the DTS system during the first water cooling stage are reduced by 0.14 °C and 0.22 °C, and their fluctuations are decreased by 41.5% and 47.0%, respectively. These results verify the reliability and feasibility of the error correction method proposed in this paper, and the method is demonstrated to provide reliable technical support for obtaining stable and high-precision DTS temperature data in concrete dams.