Study On Preprocessing Method Of Structural Deformation Monitoring Data

With the rapid development of economy,the number and scale of structural engineering construction are increasing,which puts forward higher requirements for the safety of structural engineering.With the application of technologies such as sensors and big data processing in engineering construction,structural automatic monitoring system has become an important means to ensure structural safety.The effective operation of structural automatic monitoring system is based on reliable monitoring data,and the structural monitoring site often leads to abnormal monitoring data,loss and other phenomena due to the influence of monitoring equipment,communication mode,outfield environment and other factors,which affect the reliability of structural safety state assessment.Therefore,it is of great research value to carry out the pre-processing and analysis of monitoring data for the structural automatic monitoring system.Relying on the national key research and development program project "Visual Automatic Monitoring System for Construction Safety of Urban Underground Large Spaces"(2018YFC0808706).Aiming at the problem that the noise reduction of structural deformation monitoring data does not effectively utilize the characteristics and applicability of monitoring data,this paper proposes a noise reduction model based on EEMD and wavelet analysis,which can effectively remove the noise signal in structural deformation monitoring data.First,wavelet denoising is performed on the structural deformation monitoring data,and the denoised data is decomposed with EEMD to obtain the random and trend items.Finally,the correlation analysis method is introduced to reconstruct the noise reduction signal by set the correlation coefficient threshold,which the random items are reserved larger than the threshold.The proposed method is verified with the real monitoring data measured from the subway subsidence.Compared with the traditional noise reduction method,the average noise reduction performances of the proposed method are improved by 32.62%,which indicates that the proposed model is an effective noise reduction method for structural monitoring data.Aiming at the isolated outliers existing in the structural deformation monitoring data,this paper comprehensively considers the detection effect and complexity of the anomaly detection method,adopts the improved 3σ criteria to detect the outliers,and repairs the outliers through the moving average method according to the relationship between the outliers and the monitoring data series.The simulation test results show that the outliers can be accurately identified and the outliers can be reliably repaired at the same time.Aiming at the problems of the existing continuous outlier repair methods of monitoring data,such as the model too complex and the contradiction between repair accuracy and operation speed,this paper proposed an outlier repair method for structural monitoring data based on BLS model considering the advantages of BLS model in processing time series operation speed.The proposed model is verified with the real monitoring data measured from the subway subsidence.Compared with ARIMA,ANN,SVR and DBN,the repair error of the proposed method is reduced by 26.91%,7.32%,1.99% and 4.71% respectively.The operation time of BLS model is 95,73,1980 and 2459 times of the other four comparison models.The results show that BLS model is effective in repairing continuous outliers of monitoring data.