Structural Damage Detection Based On Reconstructed Phase Space And The Visualization Of Damage Information

In this peper, based on the current problems and challenges of structural damage detectionand health monitoring, a new damage detection method, and the method of visualization ofdamage information are proposed according to the fisibility of implementation, practicalapplicability and observation. Firstly, the significance of this research field is introduced, and thecurrent structural damage detection and strucrural health monitoring methods are classified.Based on the analysis of current research situation of this field in the world, the hot researchissues, current problems, difficulties and challenges in the application of real engineeringstructures are summarized.Several matural used modal-based methods are used to identify the damages of a circulararch. Then, the influence of the damage detection results under using different number ofmeasurent points to extract the modal parameters is analized. And also the robustness of theresults under different environments such as diferent levels of noise is researched. Theadaptability of practical application of modal-based methods is analized for the comparison withthe proposed method in the following studies. The simulation results indicate that the reliabilityand accuracy of modal-based methods decline because the loss of modal information when usingless number of measurement points, even some parameters can not identify the damage at all.When the mode shapes are smeared with noise to simulate the inference of environment, theresults indicate that none of the modal-based methods can detect the damage of the arch.In order to overcome the disadvantage of the modal-based methods, a new damagedetection method based on vibration reconstructed phase space is proposed. Firstly, thedefination of the phase space and reconstructed phase space of vibration system are introduced.The methods of determination of embedding demenssion and delay time are analized. In thismethod, the technology of phase space reconstruction sterming from the study of chaotic series isintroduced to structural damage detection. Then the fisibility of this proposed method is studiedusing a one-demenssion spring oscillator. The results indicate that the reduction of stiffnessintroduces the changes on the topology of phase space, which indicates that this method is fisibleand reliable. The damage parameter is proposed based on the changes of reconstructed phasespace to detect the damage which is set as the same in modal-based analysis. The simulation andexperiment are carried out on the circular arch and the results are compared with the modal-based methods. The results indicate that the proposed method is much more sensitive thanmodal-based method, and presents good robustness to the noise, which prove that the method isfisible abd reliable. An advantage of this method is that response of only one measurement pointcan be sued to caucluate the damage parameter of this point, which proves that this parametrercan be considered as a global index to monitor the health of the structure. In order to furtherprove the reliability of this method, an experiment of damage detection and health monitoring ona6m length reinforced concrete slab with two spans is carried out. The results indicate that thismethod can succefully identify the damage of the slab. It is also indicated that the influence ofembedding dimension to the detection result is significant. The result of embedded twodimensions condition is not as good as three embedding dimensions because some damageinformation which only affetc the third embedding dimension direction is missed.Another new damage detetion method is also proposed based on the nonlinear chaoticDuffing system. The largest Lyapunov Exponent of the phase space of the output signal of thesystem is extracted and considered as the damage parameter. This method sterms from the weeksignal detection using Duffing chaotic system. The theory of this method and the method of thelargest Lyapunov Exponent calculation are introduced. In this mehod, the vibration signals areinputted into the Duffing system directly, which makes the system at the critical status become tochatic status quickly, and then the largest Lyapunov Exponent calculation is calculated. Usingthis proposed method, the damage of a simple beam is identified successfully with only onesensor. The results indicate that it is more sensitive than natural frequencies.Last, the method of hierarchical and fused visualizations of damage detection using manyindices is proposed. This method makes the damage information be displayed directly withvisual image. The process of visualization is introduced and the informations fusion method isproposed. This metod uses the difference of the colors to indicate the damage location and level.Then the damage information is mapped onto the real structure, so as to present the damagevisually and easy to be understood. Thus, it can be promoted and used in the real structuraldamage detection and health monitoring as a new display platform.