An innovative noise reduction algorithm for 5G millimeter-wave monitoring of urban bridge micro-deformation

Abstract The monitoring of urban bridge micro-deformation is crucial for ensuring the safe operation of bridge, preventing disasters, and improving construction efficiency. However, existing monitoring methods face challenges such as difficult deployment, maintenance, low precision, and low frequency, making it hard to obtain dynamic deflection data of urban bridge. In recent years, millimeter waves emitted and received by 5G base stations, with their non-contact, high-frequency, and 7 × 24 all-day advantages, have begun to be applied to the high-dynamic monitoring of urban bridge dynamic deflection. Nevertheless, 5G millimeter wave monitoring is susceptible to noise due to complex environmental factors, leading to unreliable dynamic deflection information of bridge. Therefore, based on the analysis of noise sources and propagation characteristics in the monitoring process, this paper proposes an Improved Second Order Blind Identification (I-SOBI) algorithm to enhance the accuracy of 5G millimeter wave monitoring by mitigating the issues caused by the linear correlation of original signals. The I-SOBI algorithm can recover source signals from noisy ones that exhibit nonlinear characteristics or are non-Gaussian distributed. It not only preserves the contribution of useful information from the separated signals during the reconstruction process but also achieves the goal of reducing the impact of noisy components on the reconstructed results. Furthermore, through simulation experiments and practical experiments applying 5G millimeter wave monitoring to bridge dynamic deflection, the effectiveness and feasibility of this method are verified. Compared with the traditional SOBI method, the proposed I-SOBI algorithm achieves a 4.6% improvement in RMSE and a 6.4% enhancement in PSNR. It demonstrates superior capability in preserving essential signal components while effectively suppressing high-frequency noise, resulting in a smoother and more reliable denoised signal. The research methods presented in this paper can improve the accuracy of 5G millimeter wave monitoring for urban bridge dynamic deflection, providing a reliable basis for structural health assessment.