Large-scale Structural Damage Detection Based On Blind Source Separation Of Vibration Signal

In recent years，the damage detection of the large structure is going to getting moreand more people’s attention． As the large structure will inevitably suffer from thedamage and destruction when constructing and using，and the damage and destructionoften endanger thousands of people’s life and property safety．So，how detect to thedamage status of these components as soon as possible is very important．In order tosolve this problem， people put forward the modal parameter identification method.However，the traditional modal parameter identification method have some defects to thesuitable for complex reality environment． This paper introduces a thought aboutintroducing the blind source separation algorithm into the modal parameteridentification． Based on this thought, this article has carried on the theory of theexperimental certification，complete the following works：Firstly，the paper introduces the origin of blind source separation algorithm，theresearch background and using environment in detail，and analyzes the research statusdomestic and overseas．This paper mainly introduces the Fast_ICA algorithm which isone of the most common algorithm in blind source separation algorithm．The paperanalyzes the operation steps and principles of the algorithm，tests and verifies the effectsof this algorithm through experiment to prepare for further research．Secondly，the paper expounds the principle of modal parameter identification，applying environment，merit and demerit．Studies and analyses the modal parameterequation in different environment modal，gives the internal relationship between modalparameter identification and blind source separation algorithm．The study shows thatprocessing vibration signal with blind source separation algorithm can make modal parameter identification adapt to real environment easily， and the accuracy ofidentification parameters will be improved．All the argument for the thesis provides thetheoretical support of thought．Thirdly， through several simulation experiments in Matlab under differentenvironment， different status and different prior condition， the paper verifies thefeasibility of modal parameter identification method based on blind source separation．The paper catch some characteristics of recognition accuracy and robustness by datacomparison． Simulation results show that blind source separation algorithm caneffectively meet the environment with complexity and low of quantity of information ofmodal parameter identification．Finally， the paper shows that there are still some shortcut in modal parameteridentification method based on blind source separation．The accuracy of blind sourceseparation algorithm is influenced by the own conditions and the damping of the systemand signal to noise．In order to solve these problems，there should be more study andresearch，and these problems also provide a reference for further research．...