| Global Navigation Satellite System(GNSS),with its advantages of fast,high precision and all-weather,is widely used for deformation monitoring of Bridges,landslides,buildings and structures,etc.,and has become one of the main technical methods of deformation monitoring.GNSS deformation monitoring technology includes three parts:monitoring,data processing and deformation information identification and warning;among them,monitoring is the basis,data processing is the method,deformation information identification and warning is the purpose.The key to GNSS deformation monitoring is how to accurately identify the deformation information in the monitoring data and give early warning.The identification and early warning of deformation information is an important guarantee for the safe operation of metamorphosis,which is of great significance for protecting people's life and property safety.Based on GNSS long time monitoring may occur in the process of individual gross error or outliers,will cause accumulation and(Cumulative sum,CUSUM)control chart the problem of high false alarm rate,was proposed based on the median of CUSUM control chart of GNSS deformation information identification and warning algorithm,and in view of the CUSUM control chart cannot effectively analysis the influence factors of time series of GNSS monitoring data,introduced Additive Season and Trend decomposition and breakpoint recognition algorithm(Breaks for Additive Season and Trend,BFAST),the GNSS deformation information recognition and warning method based on the improved BFAST algorithm was constructed.At the same time,considering the above method is based on the requirement that GNSS monitoring data obey normal distribution,aiming at the problem that GNSS monitoring data sometimes do not obey a specific distribution,a GNSS deformation information inspection algorithm based on local weighted regression residual was established.The main research results of this paper are as follows:(1)According to the traditional CUSUM control chart theory,the GNSS deformation information recognition and early warning algorithm based on CUSUM median control chart was established.This method adopted the method of median,combined the monitoring data,and largely eliminated the influence of individual coarse errors or outliers in the monitoring data on the detection results.Aiming at the problem that GNSS does not satisfy normal distribution,a method of non-normal data transformation was introduced.At the same time,through the measured data with simulated deformation data,the contains no gross error,the simulation of discrete gross error and gross error consecutive deformation data of relevant experiments,the experimental results showed that the method based on median CUSUM control chart can effectively suppressed the influence of noise to the deformation information,enhanced the deformation characteristics of the monitoring sequence.For the deformation sequence without gross errors,the accuracy of deformation information recognition based on median CUSUM control chart is higher than that of classical CUSUM method;for the deformation sequence with gross errors,the advantage of median-based CUSUM control chart is more obvious,and it shows good robustness,which is basically the same as that of using this method to identify the deformation sequence without gross errors.(2)BFAST is a breakpoint detection algorithm for decomposing time sequence.For the problem that CUSUM control chart cannot effectively analyze the influencing factors of time series in GNSS monitoring data,a GNSS deformation information recognition and early warning method based on improved BFAST algorithm was proposed.In the experiment,it was found that BFAST was not suitable for the case where the amplitude of GNSS coordinate sequence fluctuates greatly.In view of this,this paper improved the BFAST algorithm by combining the advantages of CUSUM control chart.At the same time,considering the fact that GNSS does not satisfy seasonality when collecting data.,optimized in the iterative step.The experimental results showed that the improved BFAST algorithm solved the problem that the original BFAST algorithm was not suitable for large amplitude fluctuations of GNSS coordinate sequences.Compared with the original BFAST algorithm,the improved BFAST method improves the reliability of the detection results.Stability,the recognition accuracy of deformation information was better.(3)In the detection of deformation information,the above method has certain conditions that the GNSS monitoring data is subject to a normal distribution.However,the GNSS signal is affected by errors such as multipath error and ionospheric delay,and the monitoring data does not obey a certain distribution.In view of this,a GNSS deformation information verification algorithm based on local enhanced regression residual was proposed.Experiments were caried out by simulating data and adding deformation information to the measured data.The results showed that no matter which method was selected for verification,the method had good ability to test deformation information.For the test of the upper,lower and positive trend data,it had obvious advantages and can accurately reflect the change trend.For the mutant data,the algorithm was suitable for continuous large offset deformation data with more than two standard deviations.The test can effectively identified the range and position of the deformation information in the GNSS monitoring data,but the recognition accuracy of the continuous small offset data below the standard deviation of the standard deviation was low.Figure[27]Table[5]Reference[87]... |
PDF Full Text Request