| By using the change of dynamic and static parameters to identify the structuraldamage is a hot research topic at present in the structure health monitoring. Using thevariable of the structure displacement and modal shape to analysis the stiffnesschanges of the structure so that identify the damage degree and location of thestructure, which has broad prospect of application in practical engineering.The model updating base on sensitivity matrix which applied to traditionalmethod of damage identification need large calculating quantity and complexiteration procedures. In this paper, using the response surface model to replace thefinite element model to establish the objective function with the structural parameterswhich are calculated by ANSYS software after damage, and use genetic algorithms toidentify the damage, this method can improve the computational efficiency. In orderto improve the identification accuracy and robustness of the algorithm, according tothe characteristics of damage identification, this paper combines genetic algorithmwith response surface and regularization processing, as well as fuzzy mathematicstheory to identify the structural damage effectively. Specific content as follows:(1) When using genetic algorithm to identify the damage it needs to do a lot ofdirect problem solving, in order to avoid the large amount of calculation whendirectly use the finite element model to calculate, this paper introduces a responsesurface equation in the genetic algorithm and uses the response surface model insteadof the finite element model to calculate.(2) The traditional genetic algorithm will fall into local optimum easily whenused to solve the structural damage objective function. This paper uses the geneticalgorithm based on gray code, it greatly improves the local search ability in theprocess of the objective function optimization and makes the damage identificationresults more accuracy.(3) Using structure parameters to identify the damage location and degree inmathematics is a process of solving the inverse problem. The inversion results occursoscillation and the damage identification will be miscalculated due to the discomfort of the inverse problem. This paper proposes regularization improvement in geneticalgorithm, this method avoids the oscillation and divergence of the results whichcaused by the discomfort of the inverse problem. This method makes the damageidentification results more accuracy.(4) The fuzzy mathematics theory is introduced in this paper, damage of thestructure the state and the structure parameters are used as input and output variablesof a fuzzy system respectively, and the response surface is instead of fuzzy rules,lastly uses genetic algorithm to find the optimal fuzzy solution and improve the noiseresistance of damage identification.According to the characteristics of the structural damage, this paper makes someimprovement on the genetic algorithm. Respectively using response surface model ofthe beam displacement and mode of vibration to establish the objective function withthe ANSYS calculated value. This paper adds the objective function with a1normregularization item and establishes genetic algorithm and fuzzy inference system toidentify the damage. The numerical example of finite element model proves that thedamage identification method proposed by this paper is effective. |
