Bridge Health Monitoring Data Cleaning Method And System Design

In recent years,the research and application of bridge health monitoring technology has garnered increasing attention from domestic and international scholars.With the increasing number of bridges being constructed in China,the utilization and standardization of bridge health monitoring systems has gradually improved.With the huge volume and spatial and temporal characteristics of bridge health monitoring data,in-depth analysis and research of these data can not only obtain the health status of bridges,but also provide status information and scientific theoretical basis for bridge design,construction,service and maintenance decisions.However,the current research on cleaning abnormal data in bridge health monitoring lacks depth.The majority of methods used to clean health monitoring data rely on traditional time-frequency analysis.This approach is only suitable for smooth signals and inadequately analyzes non-smooth,non-linear random signals,leading to substantial inaccuracies in the identification of results.Therefore,it is both imperative and urgent to explore data anomalies and cleaning methods regarding the monitoring of bridge health.Based on the requirements of the Ministry of Transport,focusing on the implementation of the key tasks of the "Outline for the Construction of a Strong Transportation Country",11in-service highway bridges in 10 provinces were identified as pilot health monitoring system construction.In this context,the present paper conducts a comprehensive study of data cleaning methods for bridge health monitoring based on the Xinjiang Guozigou Bridge Health Monitoring Project.This study integrates both domestic and international data cleaning methods for health monitoring in order to ensure accuracy and reliability.This paper provides a comprehensive summary of the existing data cleaning approaches for monitoring data both domestically and internationally.By combining the strengths of noise-complete empirical modal decomposition with wavelet algorithms,we propose a novel approach-the noisecomplete ensemble empirical modal decomposition with a wavelet threshold noise reduction algorithm.This methodology allows for effective cleaning of monitoring data with high levels of noise and provides a more accurate representation of the underlying features of the data.Furthermore,a platform for data cleaning of bridge monitoring was designed to compare the effectiveness and reliability of the proposed algorithm with mature algorithms from both domestic and international sources.This comparison was based on the cleaned data resulting from the empirical modal decomposition technique and wavelet threshold noise reduction algorithm.The purpose of this validation was to demonstrate the practicality and reliability of the proposed algorithm within the field of bridge health monitoring data cleaning.Corrections made to the original text include improving sentence structure,correcting grammatical errors and applying standard academic language protocols.The findings of this study demonstrate that the noise-complete Ensemble Empirical Mode Decomposition method,combined with the Wavelet Threshold Noise Reduction Algorithm,exhibits a high degree of suitability in effectively reducing noise in bridge health monitoring data.This conclusion suggests that this technique holds potential for practical application in bridge health monitoring systems.When confronted with non-stationary and non-linear random signals,the algorithm effectively avoids modal aliasing and misalignment of the time-frequency distribution.Furthermore,it demonstrates impressive efficiency when processing large data samples,while maintaining a reasonable root mean square error.In the face of non-stationary,non-linear random signals,it is necessary to implement noise-complete ensemble empirical mode decomposition in combination with wavelet threshold noise reduction algorithms to effectively clean the data.Such methods conform to the academic conventions of a graduate-level research paper and help to rectify any existing grammatical errors.The computational method presented in this paper provides theoretical support for the data cleaning and analysis of non-stationary and non-linear random signals.This method will facilitate the enhancement of bridge health monitoring systems.