Structural Damage Detection Based On Artificial Fish Swarm Algorithm

Structural health monitoring has a large influence on sustainable development strategyof civil engineering. As the core part of structure health monitoring, structural damagedetection (SDD) has been studied and widespread attention by scholars all over the worldin recent20years. SDD method based on dynamic test is one of the widely recognizeddetection methods. It uses dynamic theory, vibration test and analysis techniques, systemidentification technology and signal processing techniques to analyze the dynamicresponse of the structure before and after damage and to detect structural damage, damagelocation and damage degree. Although many SDD methods have been proposed in thisfield, most of them are limited to specific numerical simulation model and test models, andthere isn’t a common SDD method applied to practical engineering. Therefore, developingSDD technology is not only the rigid demand in the field of structural engineering but alsoa research hotspot and difficulty in the field of structural health monitoring.Based on the input and output of mathematical model for SDD, which is the objectiveoptimization function with constraints, the contents of this dissertation are summarized asfollows:1) Traditional modal parameters identification methods and the stochastic subspaceidentification (SSI) method are studied. Some numerical responses of the benchmark four-layer framework model are used to identify the modal parameters of the healthy structure.The results show that SSI is more accurate than the frequency response function method aswell as the peak-picking method.2) An improved global artificial fish swarm algorithm (GAFSA) has been proposed.The effect of parameters of artificial fish swarm algorithm (AFSA) on the computationperformance, such as convergence accuracy, convergence speed and computationalefficiency, has been studied. The value range and optimal combination of parameters areobtained by using three benchmark test functions. The results show that the computationperformance of GAFSA is much better than the basic AFSA.3) The applicability of AFSA method to the SDD field has been investigated. Somesymmetric and asymmetric damage conditions of both the ASCE benchmark four-layerframework and the two-layer plane frame are set respectively. The numerical simulationresults show that the proposed AFSA-based SDD method is effective and feasible for theSDD problem. 4) The step by step method of structural damage detection is proposed. Firstly, thenumber of damage elements is determined using the residual force vector method. Then thelocations of damage elements are identified using the modal strain energy ratio. Lastly, thedamage degrees of damage elements are estimated based on the AFSA-based SDD method.Some numerical simulations of a plane truss structure show that the step by step SDDmethod has good noise immunity and it can accurately detect the structural damages in thecase of incomplete mode shapes.5) A series of SDD experiments of a steel truss bridge model have been conducted inlaboratory. Firstly, the modal parameters of the model under each condition are identifiedusing the SSI method. The finite element model is then updated using the measured modalparameters. Lastly, the structural damage detection has been conducted through theupdated finite element model and the identified modal parameters under other damageconditions. The illustrated experimental results show that the SDD methods proposed inthis thesis can successfully detected the structural damages of steel truss bridge modelfabricated in laboratory.