Condition Assessment Based On Cointegration And Multivariate Control Chart For Large-span Bridges

Large-scale bridge structural health monitoring, with its real-time acquisition of bridge structure degradation condition and timely warning, can reduce the maintenance costs of bridge and prevent sudden accidents such as bridge collapse, which makes bridge structural health monitoring the research frontier in civil engineering. With the impact of environment temperature, load, etc, operating bridge structure degradation is difficult to find directly through the monitoring data. At the same time, due to the complex structure of large bridge, the theory of finite element model calculation results can’t be consistent with the bridge monitoring data, leading to fact that the safety state evaluation based on constant environment and finite element model can hardly be applied to actual bridge. Every measuring point of bridge structure has common trends in strain response and displacement response under the influence of temperature and load. According to the change of structure response statistics happened before and after the damage, the paper will use multivariate control chart to identify the structure damage by establishing separation algorithm of effects based on the co-integration theory. In this way, we can avoid the inconsistency between complex finite element modeling theoretical calculation and practical measure of bridge, which provides a strong support for practical application of bridge health monitoring in large bridge.The main content and results of the paper are as follows:① Through analyzing the main load effect kinds of the large bridge, we can simplify the various load effects into a total of three kinds, that is concentrated load, uniformly distributed load and temperature load. Based on the principle of structural mechanics, we deduce the deflection change rule among different points of simply supported beam model under three different load effects. The results show that under the same load effect, there is a linear trend between each point deflection of the structure;② Make the structural response of long-term monitoring data a time series data, use Johansen method to determine cointegration model of data, represent the structural response data eliminated from the effect through co-integration allowance data calculated by the model, and we can establish a separation method of effect based on the co-integration theory. Numerical example and the finite element model calculation example show that this method can effectively separate the effects such as temperature, vehicle load, etc.③ By an optimized analysis of parameter evaluations such as average operation chain length, average variation, smoothing parameter and control line in control chart, we can establish a bridge structure safety state assessment method which is based on multivariate control chart by using co-integration allowance data. Accordingly, we can come up with judgmental algorithm of single and several injuries through the option of sensor channel contained in cointegration calculation. The results of numerical example and finite element model example show that multivariate control chart can identify the structure injuries occurred in single and several places.④ Apply the effect of separation methods and multivariate control chart method to a continuous rigid frame bridge, and evaluate the safety of the bridge state through the long-term monitoring of deflection, strain and temperature data. We find out that the evaluation result is consistent with the artificial test result, which states that this method can be successfully applied to large bridge structure safety state evaluation.