Research On Location And Elimination Of Nonrandom Errors In Deformation Monitoring

In recent years,Earthquakes,debris flows,dam deformation,bridge collapse,land subsidence caused by high-rise buildings,etc.More and more attention has been paid to the research of deformation monitoring.In the actual deformation monitoring measurement.Due to human factors,instrumental factors,natural factors,etc.,the observations include gross errors and systematic errors.The existence of these non-random errors seriously affects the quality of the observed data.The accuracy of the measured data results cannot be met,and the prediction cannot be made very well.In order to improve the accuracy of deformation monitoring.The handling of non-random errors is also particularly important.In many cases.Due to the complexity of the reasons for non-random errors.It is difficult to locate and eliminate it by means of observation or classical adjustment methods.This article is aimed at large sample data in deformation monitoring.That is,when there are more observations.System error and gross error are processed by residual observation method,Wright criterion and statistical test method.For small sample data,that is,when the observed data is small,Systematic errors and gross errors are processed using non-statistical methods such as grey diagnosis and fuzzy diagnosis.Combined with the engineering examples of deformation observation,the statistical methods and non-statistical methods are used to verify and compare the results of non-random errors under different conditions.The paper mainly includes the following research contents:(1)Analyzing the theory of deformation monitoring and monitoring and measurement methods systematically.Comparing the characteristics and differences between conventional geodetic methods,baseline measurement methods,and GPS deformation monitoring methods.The sources and characteristics of non-random errors in deformation monitoring are discussed.System errors and gross errors were preprocessed.(2)Systematic errors for large sample data in deformation monitoring.Calculate the residual value based on the observed data by the residual calculation formula.Judging and diagnosing by residual data or by plotting a residual map;For gross errors in large sample data,Positioning and culling by combining Wright criteria and statistical tests.The method is verified by an example of measured railway deformation monitoring.(3)Systematic errors for small sample data in deformation monitoring,calculating the degree of gray correlation between data based on the relationship change between the two systems.Then diagnose the magnitude of the systematic error by determining the magnitude of the gray correlation degree.Then use fuzzy mathematics,information entropy diagnosis method,Gross data positioning and culling of data sources.The method was verified by the example of settlement monitoring of foundation pit engineering.The speed and accuracy of random error observation data processing in deformation monitoring are improved.