Damage Identification Algorithms Considering Temperature Variation For Civil Structural Health Monitoring

The safety of civil engineering structures is closely related people’s life and property. In order to detect structural damage or degeneration, maximize structural service life and minimise life-circle cost, it is necessary to apply structural health monitoring (SHM) system in high-rise buildings and long-span bridges. Structural damage identification algorithms are the key of the SHM system, it has been intensively researched during the past thirty years; however, the complexity of practical projects deteriorates the effectiveness of these algorithms. In recent years, identification algorithms considering environmental effects and operational conditions become the research hotspots. Damage identification methods considering environmental and operational variables make a great progress in damage detection techniques of SHM. The work done in this dissertation was listed as follows:(1) A thorough description of the important status of SHM in the civil engineering area, and a general review about damage identification methods were presented; particularly, vibration-based methods were reviewed in detail. Further more, the negetive effects of various environmental factors and operational conditions on structural parameters were summarized, and a comprehensive review was presented about two kinds of damage identification methods considering these factors in recent years.(2) Method of element modal stain damage index (EMSDI) was applied in the damage identification of transmission towers. EMSDI was calculated from structural mode shapes which were less affected by environmental and operational conditions. EMSDI was formulated step by step and was tested in a finite element model and a model experiment.(3) Blind source separation (BSS), specifically, AMUSE and FastICA algorithms, was introduced to remove the environmental effects on modal frequencies. Based on the assumption that reasons causing the change of frequency could be categorized into the damage source and environment source, the accuracy of damage identification can be highly improved by eleminating the retrieved environment source. The numerical simulation shows that the occurrence of type I errors (false positives) and type II errors (false negetive) is reduced by applying BSS algorithm.(4) Cointegration-based damage identification algorithm is proposed for the removal of environmental and operational variables. The group of neighboured displacement time series was used to locate the damage through comparison of magnitudes and directions of jumped cointegrated residuals. The validaty of the algorithm was demonstrated by a finite element model, showing that cointegrated residuals provide a reliable tool for damage alarm with the combination of control chart. In addition, autoregression filter can be employed to improve the performance when nonlinear environmental effect is considered.