Songtao Xue

Identification of Structural Systems Using Particle Swarm Optimization

Abstract Particle swarm optimization (PSO) is a new heuristic method that has yielded promising results for solving complex optimization problems. Its advantages are a simple structure, ease of use, quality of solution, and robustness. This paper utilizes the PSO algorithm for parameter estimation of structural systems, which could be formulated as a multi-modal numerical optimization problem with high dimension. Simulation results for identifying the parameters of multiple degree-of-freedom (DOF) linear and nonlinear structural systems are presented to demonstrate the effectiveness of the proposed method.

Parameter estimation using a CLPSO strategy

As a novel evolutionary computation technique, particle swarm optimization (PSO) has attracted much attention and wide applications for solving complex optimization problems in different fields mainly for various continuous optimization problems. However, it may easily get trapped in a local optimum when solving complex multimodal problems. This paper utilizes an improved PSO by incorporating a comprehensive learning strategy into original PSO to discourage premature convergence, namely CLPSO strategy to estimate parameters of structural systems, which could be formulated as a multi-modal optimization problem with high dimension. Simulation results for identifying the parameters of a structural system under conditions including limited output data and no prior knowledge of mass, damping, or stiffness are presented to demonstrate the effectiveness of the proposed method.

Application of a grey control system model to structural damage identification

Development of a health monitoring system for construction has become an important task for structural damage identification. Since the building structure has many uncertain factors, the method of using modal analysis to identify the structural damage is of low accuracy. A method applying a grey system to the structural damage identification is presented. The grey system for the frequency variance rate and the stiffness change is established with the first-order single-argument grey system (GM(1, 1)). GM(1, 1) prediction is used to reflect the global function of the structural dynamic fingerprints and find the relation of the frequency variance rate and the stiffness change. Vibration tests for frame structures were carried out, with many cases considered, including single-damage and multi-damage ones with different degrees and locations. The results show that for shear buildings, the damage degree and location can be determined by using the grey system and measuring the frequency change.

Structural Damage Detection Using Auxiliary Particle Filtering Method

Structural damage identification is an important objective of health monitoring for civil infrastructures. Frequently, damage to a structure may be reflected by a change of some system parameters, such as a degradation of the stiffness. In this paper, an auxiliary particle filtering (APF) method is applied to track a dynamic system with sudden parameter changes. In the APF, the importance density is proposed as a mixture density that depends upon the past state and the most recent observations, and hence which has a good time-tracking ability that is more suitable for tracking the nonstationary system than the conventional particle filters. Simulation results for tracking the sudden parameter changes of nonlinear hysteretic structures are presented to demonstrate the application and effectiveness of the proposed technique in detecting the structural damages.Structural damage identification is an important objective of health monitoring for civil infrastructures. Frequently, damage to a structure may be reflected by a change of some system parameters, such as a degradation of the stiffness. In this paper, an auxiliary particle filtering (APF) method is applied to track a dynamic system with sudden parameter changes. In the APF, the importance density is proposed as a mixture density that depends upon the past state and the most recent observations, and hence which has a good time-tracking ability that is more suitable for tracking the nonstationary system than the conventional particle filters. Simulation results for tracking the sudden parameter changes of nonlinear hysteretic structures are presented to demonstrate the application and effectiveness of the proposed technique in detecting the structural damages.

Dynamics of Real Structure in Fresh, Damaged and Reinforced States in Comparison with Shake Table and Simulation Models

Abstract The dynamics of a real three-story structure were studied based on changes of its natural frequency when beams and braces were removed, to simulate damage, and returned, to simulate reinforcement. Total 81 steps simulating different structural states were adopted for these testes. For comparison, laboratory shake table experiments were performed with 1/20-scale models, for the same 81 steps. In addition, numerical simulations of the real structure were also carried out over the same 81 steps for comparison. The change in natural frequency for the three methods, together with the influence of temperature and humidity, showed interesting tendencies, which prove important and meaningful for the development of structural health monitoring systems using dynamic data.

Multi-objective differential evolution for truss design optimization with epistemic uncertainty:

A robust multi-objective optimization method for truss optimum design is presented. In the robust design, materials and loads are assumed to be affected by epistemic uncertainties (imprecise or lack of knowledge). Uncertainty quantification using evidence theory in optimum design subject to epistemic uncertainty is undertaken. In addition to a functional objective, an evidence-based plausibility measure of failure of constraint satisfaction is minimized to formulate the robust design into a multi-objective optimization problem. In order to alleviate the computational difficulties in the evidence theory-based uncertainty quantification analysis, a combined strategy of differential evolution-based interval optimization method and parallel computing technique is proposed. A population-based multi-objective differential evolution optimization algorithm is designed for searching robust Pareto front. Two truss structures with shape and sizing optimum design problems are presented to demonstrate the effectiveness and applicability of the proposed method.

Dynamic Properties of Conventional Beam-Column Timber Structure Under Successive Damage

Abstract A full-scale, three-story conventional beam-column timber structure was tested in-situ to investigate the changes of natural frequencies due to progressing artificial damage. A total of 45 test steps were adopted in the execution of this experiment and analysis. Columns were cut by saw and/or totally removed at each testing step in order to simulate different levels of damage. A three-dimensional finite element (FE) model of the test structure was established and applied to predict its natural frequency in each of the 45 test steps. Both the FE simulation and the testing results show similar tendencies in most test cases, however they slightly deviate from each other in some particular test cases. Damage sensitivity, as well as the influence of temperature and humidity on the natural frequency was also examined. The results and conclusions from this study can benefit the emerging research field of structural health monitoring.

An Adaptive Multi-objective Immune Algorithm for Optimal Design of Truss Structures

In this paper, an adaptive immune clone selection algorithm for multi-objective optimization (AICSAMO) is proposed. A novel adaptive polynomial mutation operator with dynamic mutation probability is employed in AICSAMO. This adaptive mutation operator executes a rapid global search at the earlier stage of the algorithm and a fine-tuning search at the later stage of the algorithm, which adopts generation-dependent parameters to improve the convergence speed and global optimum searching ability. The effectiveness of AICSAMO is evaluated through the truss sizing and shape optimization problems of a 10-bar plane truss and a 25-bar space truss. According to the comparison of AICSAMO with various multi-objective optimization algorithms developed recently, the simulation results illustrate that AICSAMO has remarkable performance in finding a wider spread of optimal solutions and in maintaining better uniformity of the solutions with better convergence.

Performance Study of an Eight-story Steel Building Equipped with Oil Dampers Damaged During the 2011 Great East Japan Earthquake Part 1: Structural Identification and Damage Reasoning

Abstract Oil dampers installed on the first floor of an eight-story steel building were completely destroyed during the 2011 Great East Japan Earthquake. It is believed to be the first time in the world that real oil dampers in service failed due to earthquakes. Before this failure event, the actual performance of buildings that use oil dampers during catastrophic earthquakes has never been verified. Investigating the cause of the damage of the oil dampers is thus necessary and urgent. In this paper, a comprehensive identification was conducted to rebuild the numerical model of this damped structure equipped with/without damaged oil dampers using the measurement data of the installed monitoring system. Furthermore, the damage process of the oil dampers was postulated based on the identification and simulation results. The limit states of the oil dampers were studied. Based on the damages of the dampers and connection, the oil dampers experienced the displacement limit state when the allowable displacement limit was surpassed and the central cylinder pushed against the abutment. The insufficient stroke limit is the main cause of the collision between the damper and the abutment on the floor, which finally led to the failure of the oil dampers.

Usage of Comprehensive Learning Particle Swarm Optimization for Parameter Identification of Structural System

This paper introduces a novel swarm intelligence based algorithm named comprehensive learning particle swarm optimization (CLPSO) to identify parameters of structural systems, which could be formulated as a multi-modal numerical optimization problem with high dimension. With the new strategy in this variant of particle swarm optimization (PSO), historical best information for all other particles is used to update a particles velocity. This means that the particles have more exemplars to learn from, as well as have a larger potential space to fly, avoiding premature convergence. Simulation results for identifying the parameters of a five degree-of-freedom (DOF) structural system under conditions including limited output data, noise polluted signals, and no prior knowledge of mass, damping, or stiffness are presented to demonstrate improved estimation of these parameters by the CLPSO when compared with those obtained from standard PSO. In addition, the efficiency and applicability of the proposed method are experimentally examined by a twelve-story shear building shaking table model.