| With the rapid development of the structural analysis theory, construction techniques and building materials on bridges, the span is increasing and structure tends to be lightweight. Therefore, not only rigorous calculation and construction technology are required, but also higher standard for security operation of the finished bridge. Health monitoring and safety assessment on the large span structure is of great significance. In order to achieve the aim of health monitoring, the structure responses are collected through sensors in a long period, and damage sensitive features are extracted from response signals, then judge whether the structure is abnormal. When damage detection method based on the structural dynamic parameters is applied to simulant signals, the detection result may be good. However, the method in the application of the actual structure and the measured signal is difficult to obtain good results. The change in operating and environmental condition is one of the important factors. Effect of environmental factors on damage indices sometimes may cover the change of damage indices caused by real structure damage, and health monitoring should not ignore the influence of environmental factors. The environmental effect on damage indices has become a hot and difficult point in the research field.In the thesis, effect of ambient temperature change on long term performance of large span suspension bridge, especially temperature effect on dynamic strain measured data, were analyzed based on the engineering background of Baling River Bridge in Guizhou. The main tasks and conclusions are as follows:1. Through the long term measured data of Baling River Bridge, temperature effect on deflection of main girder, inclination of tower, displacement of expansion joints and force of the hanger and the cable strand were analyzed. Temperature changes of tower and structure in middle cross section of main span were understood. The results showed that the data reflecting the change of structural geometric configuration (deflection, inclination and displacement) and hanger force had obvious correlation with temperature.2. Dynamic strains of the active strain gauges fixed on chord of steel truss at mid span of suspension bridge and inactive strain gauges, that didn’t sense strains produced by stress but temperature variation induced strains, were measured for a long time. Temperature effect components of strain history showed trend form that experience in a long time. While the components of dynamic load effect, such as vehicle, presented impulse form that lasted shortly. Signal decomposition method was proposed to separate the information in strain data of the active gauges into temperature effect component and live-load effect component. According to power distribution of signals of the active and inactive strain gauges, the cut-off frequency was obtained for signal decomposition. Then, the measured strain data of active strain gauges were separated into the slow-varying component and fast-varying component by analytical mode decomposition method based on the cut-off frequency. It was shown that strain history fluctuation (slow-varying component) induced by varying temperature was extracted. Through the method, all of the gauges strain and stress-induced strain under temperature variation could be effectively eliminated. Via analysis of fast-varying component using rain-flow counting method, fast-varying component properly retained the strain information produced by dynamic load.3. Maximum stress amplitudes of measuring points were obtained through the extracted fast-varying component during the test. According to amplitudes, it’s concluded that maximum vehicle loads of test period in January and April on Baling River Bridge showed little difference. After eliminating the temperature variation-induced gauge strain from slow-varying component, the remaining component reflected the change of structure temperature stress of measuring points. In April, a larger daily temperature difference leaded to much higher temperature stress of measuring points than January and temperature stress produced by daily temperature difference was greater than maximum stress amplitude induced by vehicle load.4. Correlation function of dynamic strain was defined. It was verified that under white noise excitation, the correlation function amplitude vector (Vcf) of strain history was only related to the natural frequency, mode shape and damping ratio of structure. After Vcf being normalized, it showed a fixed ratio shape. This conclusion was proved by studying temperature effect on correlation function of dynamic strain based on test data.5. Based on principal component analysis, a method for eliminating the effect of environmental temperature was proposed. The principal component analysis was applied to the measured strain response, and the feature subspace mainly reflecting the temperature information was obtained. Through projecting the measured strain history into feature subspace of temperature and data reconstruction, temperature effect component was extracted. Then, temperature effect on strain signals could be removed. The analysis results of test data showed that the first principal component indicated most of the volatility of information. Through reconstruction of measured data with eigenvector of the first principal component, the temperature effect components of strain history were effectively extracted.6. In different time period, the shapes of Vcf based on measured data were distinct. After temperature effect component being eliminated from test data, the shapes of Vcf performed consistent. It was showed that temperature effect component in test data leaded to various shapes of Vcf. Therefore, ignoring the temperature effect on the Vcf of strain history was prone to get the false judgment on the real operation condition of structure. |
