Civil structures suffer deterioration either for years of service, deficiency due to environmental factors or damages caused by factors such as earthquakes, winds, impact loads, and cyclical loads. When a structure ages, it is necessary to know its state of health and make a decision of maintenance or replacement. When a structure such as a bridge or building is subjected to destructive environmental forces, determining its state of health becomes a priority since its recovery is urgently required to function normally. Structural Health Monitoring (SHM) is a technology that aims to prevent the collapse of structures and loss of human life through early diagnosis of the health status of a structure. There are a large number of damage detection methods that can be classified into (1) non-destructive testing methods, (2) dynamic characteristics-based damage detection methods, (3) dynamic response-based, (4) multi-scale damage detection method and (5) damage detection methods with consideration of uncertainties. In this work, it is implemented synchrosqueezed wavelet transform (SWT), which can be classified as a methods based on the dynamic response. To validate the robustness of the method it is identified first, the natural frequencies of the Benchmark Phase I without damage, which consists of a steel structure of 4-story [Formula: see text] bay 3D steel frame structure subjected to ambient vibrations. Subsequently, some damage patterns are validated according to IASC-ASCE SHM Task Group. The results obtained in the identification of natural frequencies are compared with those reported in literature. SWT was efficient, presenting a minimum error of 0.12[Formula: see text] and a maximum of 3.06[Formula: see text] in the identification of natural frequencies about the AISCE-ASCE group model. SWT overcomes some other damage detection methods, which are deficient in the identification of closely spaced frequencies, commonly present in many civil structures due to symmetric geometry or similar physical properties in different directions.
Sleep Spindle Detection Using RUSBoost and Synchrosqueezed Wavelet Transform
