An integrated study that explores the capability of Time-Frequency Analysis combined with another technique of signal analysis and system identification for damage detection purposes using only output signals is presented in this work.;The necessary background theory of Time-Frequency Analysis is presented, some definitions of damage and general approach employed in this research is given. The problems associated with the selection of a best-performance Time-Frequency Representation (TFR) for structural damage detection in civil engineering structures are discussed and a new multicriteria measure for an objective and user-independent selection of Time-Frequency Representation is developed. An evaluation of the proposed multicriteria method to select a TFR from several Linear and Quadratic TFRs is shown using synthetic ambient vibration signals and real strong motion records.;A new methodology for system identification and damage detection using output data from Ambient Vibration records is developed. This methodology is based on direct application of a proposed Mean Time-Frequency Representation (MTFR) and the Frequency Domain Decomposition (FDD), this last technique was developed by Brincker et. al. [124]. The methodology proposed in this work has been tested using synthetic and real ambient vibration signals. Several real case studies using ambient vibration data from structures are shown.;The damage detection using Time-Frequency Representations of output records of strong events are addressed, in this regard a new algorithm for instantaneous frequency tracking from TFR maps is proposed and a mathematical expression to link instantaneous frequency changes with damping and structural stiffness loss is also developed. The expression and algorithms have been tested for damage analysis of non-linear SDOF and MDOF using numerical data.;A Graphical User Interface (GUI) program named "Time-Frequency Damage Detection Program - TFDDP" has been developed using MatLab RTM[125] language; the developed algorithm has been published as a GNU -- Licensed Open File, and can be downloaded from the Puerto Rico Strong Motion Program web page at the internet site: http://prsmp.uprm.edu/ or contacting to author of this research on the e-mail: leocano-s hotmail.com
展开▼
