A Study On Reliability Method For Structural Damage Identification Based On Dynamic Fingerprints

Structural damage identification is an important component of structural health monitoring system, and is also currently a research area of much interest and importance in field of structural health monitoring system. With the traditional structural damage identification methods, structural damages are identified using deterministic data for the structural model and test results of static and dynamic responses, and the identification outcome is treated as deterministic. However, the construction and operation of a structure and the test data collection are subject to the influence of many uncertain factors, making the model data and experiment data both random. Furthermore, the randomness of parameters will lead to the probability deviations of structural damage identification results, causing possible misjudgment of damage and serious affect on the reliability of damage identification results. When the reliability of the damage identification results is too low, the damage identification methods will fail and become useless.In this dissertation, the study on the dynamic fingerprint-based structural damage identification reliability methods is developed on the basis of structural reliability theory and identification principle of dynamic fingerprint methods. The following research results are achieved in this dissertation:(1) Concepts and composition of structural health monitoring system together with structural damage identification methods are briefly introduced, and then various dynamic fingerprint methods of structural damage identification are systematically summarized, including identification principles of various common dynamic fingerprints as well as advantages and disadvantages of them. In addition, probability analysis methods of structural identification theory are comprehensive reviewed.(2) Some basic concepts of structural reliability are presented in detail, such as reliability, degree of reliability, limit state, performance function, failure probability and reliability index. Moreover, some types of conventional reliability analysis methods are described and their advantages and disadvantages are commented, including first-order second-moment method, JC method, reliability calculation method with normal random variables, response surface method and Monte-Carlo method.(3) Three dynamic fingerprint-based damage identification methods are discussed, including Average Curvature Mode Difference Damage Factor (ACMDDF), Average Modal Strain Energy Change Rate (AMSECR) and Flexibility Difference Curvature Damage Factor (FDCDF). Then deterministic methods of dynamic fingerprint-based structural damage identification are systematically presented. In addition, deterministic method for establishing the relationship between damage extents and dynamic fingerprints based on the finite element method and spline fitting is proposed emphatically and an iterative algorithm of multi damage extents deterministic identification based on dynamic fingerprint is developed. Furthermore, the validity and practicality of the method is shown with a numerical example of simply-supported girder bridge.(4) The concept of structural damage identification reliability is clearly defined for the case when the variations of structural parameters and testing parameters are considered. Based on this concept and structural reliability methods consisted of response surface method and Monte-Carlo method as well as deterministic method of dynamic fingerprint-based structural damage identification, a systematic set of dynamic fingerprint-based structural damage identification methods are proposed and implemented, which consist of calculation method of structural damage identification reliability and reliability-based structural damage identification method. Furthermore, a new approach in view of probability analysis is proposed to dynamic fingerprints anti-noise evaluation and random parameters sensitivity analysis. With simply-supported girder bridge as an example, the comprehensive damage statistic analysis are conducted using the proposed methods, and the interaction between different factors such as damage extent, damage identification reliability, damage identification accuracy and noise level are investigated for different dynamic fingerprints. The anti-noise performance of studied dynamic fingerprints and the sensitivity of damage identification results to various random parameters are also analyzed. The validity and practicality of proposed structural damage identification reliability method is validated and the valuable conclusions are made.(5) Characteristics and mechanisms of structural damages in long-span arch bridge are summarized and analyzed. Then the comprehensive probability analysis of main-span arch rib damage of a long-span arch bridge with a 428m main span length is performed using the proposed dynamic fingerprint-based structural damage identification reliability methods. Under the noise environment, the identification reliability of main-span arch rib damage for a given damage accuracy and the distribution intervals of main-span arch rib damage identification results for a given damage identification reliability are calculated. The validity and practicality of proposed methods on the application of engineering is validated. Furthermore, on the above basis, the dynamic fingerprint identification capability of main-span arch rib and main-span deck are evaluated.The major innovative points of this study are as follows: (1) A deterministic method is developed to relate the damage extents and dynamic fingerprints based on the finite element method and spline fitting, and an iterative algorithm is proposed for identifying extents of multiple damages in a structure based on dynamic fingerprint.(2) The concept of structural damage identification reliability under noise environment is proposed and properly defined. Systematically proposing dynamic fingerprint-based structural damage identification reliability method on the basis of structural reliability methods consisted of response surface method and Monte-Carlo method and with the deterministic structural damage identification dynamic fingerprint method as implementation tools to form the response surface function.(3) A new approach in view of probability analysis is proposed for the dynamic fingerprints anti-noise evaluation and random parameters sensitivity analysis. The identification capability of different dynamic fingerprints and the influence of various parameter noises on identification results are evaluated in a more objective way.