Multi-index Joint Warning Of Long-span Rail-cum-road Cable-stayed Bridge Based On Displacement Monitoring

With the rapid development of large span Bridges in our country and people’s demand for safety and comfort,health monitoring system has become an indispensable device to ensure the safe operation of Bridges.The purpose of bridge health monitoring is to ensure the safe operation of the bridge,and the basis of judging whether the components of the bridge work normally is the data collected by the sensor.By analyzing the data,the internal relationship between them can be obtained,and the alarm can be given when the bridge performance is abnormal.Therefore,the accuracy of data processing and analysis and the authenticity of system warning when damage is particularly important.However,at present,bridge performance early warning can only be judged by a single index.Due to the disadvantages of single variable early warning,it may lead to more false alarms.Therefore,this thesis proposed joint early warning with similar variables,and used the correlation between variables to backinfer the reliability of the alarm,effectively reducing the probability of false alarm in the system.This thesis takes a long-span railway bridge as the research background and longitudinal displacement monitoring data of the bridge as the research object.Specific work is as follows:(1)The common faults of sensors in the health monitoring system are classified,and the basic principle of principal component analysis is applied to establish the sensor fault diagnosis method;Then,the cumulative residual contribution rate method is used to locate the fault location.Since the reason for the abnormal data of the sensor is not the sensor fault,other factors may also cause the abnormal data of the sensor.Based on this possibility,this thesis uses the method of information fusion to carry out correlation analysis on the reasons for the abnormal data of the sensor.(2)The monitoring data was processed and analyzed to obtain the correlation between external factors,structural response and similar structural response.Firstly,the empirical mode decomposition method is applied to smooth the original monitoring signal,so as to minimize the influence of the external environment and obtain more representative data.Secondly,the grey causal model among the influence factors is established for the processed data,and the causal relationship is obtained.Finally,a causal model among longitudinal displacement monitoring indexes is established to lay the foundation for joint early warning.In this thesis,the grey causality model,linear regression model and BP network data fitting are compared.The results show that the grey causality model is better than other methods in data fitting,and the results obtained are more convincing.(3)Based on the grey model between the displacement of the bearing and the displacement of the expansion joint,a joint warning mechanism is established.Using the method of mean control chart,the warning limit was set,and the data was divided into two parts: training samples and test samples.The test samples were subjected to a certain degree of damage as different working conditions for verification.Because of the different position of the support,and the degree of association between the expansion joint is different,so respectively analyzed the different position of the support in different working conditions and the expansion joint alarm synchronization and correlation;Then,the apriori algorithm is applied to verify the correlation of alarm.The results show that the joint early-warning of bearing and expansion joint has a high support rate and confidence rate.