A Method for Separating Long-Term Deflection Effects of Bridges

Abstract

The invention discloses a separation method for a long-term deflection effect of a bridge, and belongs to the technical field of bridge health monitoring. According to the characteristics of each element of the deflection of the bridge, a non-linear relationship of temperature and a temperature deflection effect is built by the utilization of a fitting model of an M-LS-SVM. The training steps for the fitting model of the M-LS-SVM are as follows: temperature data of the bridge are decomposed in an empirical mode, and the decomposed temperature data serve as input signals of the M-LS-SVM; sample data are classified through a subtractive clustering algorithm; the classified sample data train each LS-SVM sub-model, and then the LS-SVM sub-models are built; predicted values of all the models are synthesized through a principle component regression method, data dependency among local systems is removed, robustness and generalization ability of the systems are improved, and finally an M-LS-SVM model is formed. A fitted value, obtained by the separation method, of the long-term deflection is high in precision.

Description

technical field [0001] The invention relates to a method for separating long-term deflection effects of bridges, in particular to a method for separating long-term deflection effects of bridges based on multiple least square support vector machines (M-LS-SVM), and belongs to the technical field of bridge health monitoring. Background technique [0002] The deflection of bridges is one of the important indicators reflecting the safety of bridge structures. Using various deflection measurement methods to obtain abnormal change signals of structural performance and early detection of safety hazards in large structures has become the research direction of civil structures. With continuous progress, it is still not an easy task to accurately predict the long-term deflection of long-span prestressed concrete. [0003] The signals obtained by long-term health monitoring often contain the combined effects of various factors, such as environmental effects, load effects, and concrete ...

AI Technical Summary

Problems solved by technology

[0002] The deflection of bridges is one of the important indicators reflecting the safety of bridge structures. Using various deflection measurement methods to obtain abnormal change signals of structural performance and early detection of safety hazards in large structures has become the research direction of civil structures. Progress continues to be made, and accurate prediction of long-term deflection in prestressed concrete with long spans is still not an easy task

[0003] The signals obtained by long-term health monitoring often contain the combined effects of various factors, such as environmental effects, load effects, and concrete shrinkage and creep effects. Studies on the measured displacement and strain of large structures show that temperature is the most important factor affecting signal changes. The main factor is that the damage signal of the structure is often "overwhelmed" by it, so that it is impossible to make a correct evaluation of the safety state of the structure directly based on the measured signal

Images