Please use this identifier to cite or link to this item:
Title: Response covariance-based multi-sensing damage detection of civil structures
Authors: Lin, Jianfu
Degree: Ph.D.
Issue Date: 2018
Abstract: Structural deterioration with time is inevitable once civil structures are built, for they are subjected to harsh environment and extreme events, such as strong winds and severe earthquakes. The functionality and safety of civil structures during their service time become a vital issue. Therefore, structural health monitoring (SHM) techniques have been developed to monitor structural deterioration and detect structural damage, if any, for the functionality and safety of structures. Vibration-based structural damage detection methods, as a significant part of SHM, have been developed accordingly in past decades. However, when they are applied to civil structures, vibration-based damage detection methods encounter a few major difficulties, such as the less sensitivity of damage index, uncertainties in modelling and measurement, the number and type of sensors and their spatial location, and damage detection algorithm and procedure. Many damage indexes used in vibration-based damage detection methods, such as natural frequencies, are not sensitive to local damage of a civil structure. Different types of sensors are often used in an SHM system for a civil structure to measure both global and local structural responses, but multi-sensing information has not been used effectively for local damage detection. This thesis thus proposes a response covariance-based multi-sensing damage index in the time-domain and the associated sensitivity-based damage detection method. The feasibility and accuracy of the proposed damage index and damage detection method are investigated through numerical studies on an overhanging beam model. The results show that the proposed damage index is sensitive to structural parameter change but insensitive to measurement noise and that the proposed damage detection method can effectively use multi-sensing information for local damage detection. A civil structure often consists of hundreds of structural members and joints, but the number of sensors installed in the structure is always limited. It is quite possible that the local damage may not be covered by the deployed sensors and its location may even not be accessible. Therefore, sensors shall be optimally placed in a structure so that the sensing information from the sensors can be used for effective damage detection. This thesis first proposes a response covariance-based optimal sensor placement method with a single objective function and a single type of sensors. The single objective function is actually formed by using a weighting factor to combine the two objective functions of response covariance sensitivity and response independence. Numerical studies are conducted to investigate the feasibility and effectiveness of the proposed method via a five bays three-dimensional frame structure. It is found that the acceleration responses often contain higher kinetic energy in higher-order vibrational modes for global structure information, displacement responses contain more kinetic energy in lower-order vibrational modes for global structure information, and strain responses are only sensitive to local damage only near the sensor locations. Therefore, a structural damage detection-oriented multi-type sensor placement method with multi-objective optimization is further developed in this thesis. The multi-objective optimization problem is formed by directly using the two covariance-related objective functions, and the non-dominated sorting genetic algorithm (NSGA)-II is adopted to find the solution for the optimal multi-type sensor placement to achieve the best structural damage detection. The proposed method is finally applied to a nine-bay three-dimensional frame structure numerically and experimentally. Both numerical and experimental studies show that the optimal multi-type sensor placement determined by the proposed method can avoid redundant sensors and provide satisfactory results for structural damage detection.
When the proposed covariance-based multi-type sensor placement method and the associated damage detection methods are applied to a large and complex civil structure, the obstacles exhibit. The global stiffness matrix, modal parameters, and dynamic responses are less sensitive to local damage of a large structure compared with a small structure. The one-stage damage detection is inaccurate and sometimes impossible due to too many unknown damage parameters and seriously ill-conditioned inversed problem for a large structure. Therefore, a covariance-based multi-stage damage detection strategy incorporating with a multi-scale finite element (FE) model is proposed for the damage detection of a large structure. In a multi-scale FE model, local detailed FE models using shell/solid elements and a global FE model using beam elements are integrated. The multi-stage damage detection is characterized by a few stages of different damage detection levels. For instance, the first stage is to detect the existence of damage and/or the location of damage, the second stage is to detect the damage-affected members in the identified damage location, and the final stage is to identify the damage source and quantify the damage severity. The proposed covariance-based multi-stage damage detection method is numerically and experimentally examined for its feasibility and effectiveness by using a testbed model of a high-voltage power transmission tower. Both numerical and experimental results manifest that the multi-stage detection method can effectively identify the damage in a joint due to bolt loosening and can even provide information deep down to the damage of bolts when a local detailed FE model is incorporated. The damage index and damage detection methods, including the response covariance-based multi-sensing damage index, the response covariance-based multi-objective single-type and/or multi-type optimal sensor placements for damage detection, and the multi-stage damage detection strategy, presented in this thesis could conquer some obstacles, if not all, in the damage identification of large civil structures.
Subjects: Hong Kong Polytechnic University -- Dissertations
Structural health monitoring
Structural analysis (Engineering)
Pages: xxvii, 221 pages : color illustrations
Appears in Collections:Thesis

Show full item record

Page views

Last Week
Last month
Citations as of May 28, 2023

Google ScholarTM


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.