A method for implementation of damage detection algorithms for civil structural health monitoring systems.

摘要

The field of structural health monitoring for civil structures is of great interest to the academic community, and increasingly to industry as well. Many SHM systems have begun to appear on real structures around the world. These systems provide remote access to many types of measurements, using many types of sensors. A typical result obtained for these systems may be to obtain stress and strain values, direct load measurements, corrosion levels or to validate an analytical model. However, of the systems that make vibration measurements, few are using that data for condition assessment. An important tool to aid in the interpretation of vibration data for condition assessment is a damage detection algorithm. These algorithms are not being used in those systems; there are many reasons for this, particularly related to the complexity of civil structures and the nature of real data. This thesis presents a method for implementation of those algorithms into a SHM system, considering the performance of the algorithms and some of the effects of real data.; The implementation method has two main parts: development of a damage detection methodology (DDM) to be used in the system, and development of a way to evaluate the performance of the DDM if it were applied to a real structure. The DDM is created as a framework from within which available damage detection algorithms can be implemented into the SHM system. The structure of the DDM is that it uses multiple damage detection algorithms simultaneously. This is done acknowledging the limitations of available algorithms and attempts to improve accuracy and robustness by combining them to exploit their individual advantages. The evaluation of the DDM is accomplished by creating a set of calibrated damage simulations (CDS). Ambient vibration signals are simulated using a FEM, and various aspects of the simulation are calibrated using real vibration signals obtained from an ambient vibration test. The goals of the implementation method are to obtain an evaluated and modified DDM, and the optimal layout of the sensors for the best performance of the DDM.; The CDS method was developed and illustrated using a 3D frame example. It was shown that the method could produce simulations that were good representations of real data. The DDM was developed using two simple examples, a 2D truss and a 3D frame. The DDM incorporated three damage detection techniques to create its final predictions. It was shown that the combined DDM provided better predictions than any of the techniques individually. The developed CDS and DDM were then applied to two real building case studies. Through the application of the implementation method, it was found that the DDM as developed with the 3D frame example was not directly applicable to the real building cases. It was found that the DDM might work only under certain conditions, such as modifying the finite element model used for damage detection and increasing the number of measured points in the SHM system. In addition an optimization scheme for the location of the sensors was proposed. These results demonstrated the benefit of applying the implementation method.