Structural Modal Analysis And Damage Diagnosis Based On CP-BSS Combined With Improved SVM

Ocean engineering structures would be operating in harsh marine environment for a long time and be affected by various factors.Intelligent monitoring technology uses artificial intelligence algorithm based on structural health monitoring to process monitoring signals to invert the damage status of the structure,or uses various information to evaluate the overall safety of the structure.Based on the research and development of structural identification theory,aiming at the current status and problems of parameter identification and damage identification,this paper introduces a novel Blind Source Separation,namely Complexity Pursuit.CP-BSS can integrate signal statistical characteristics and time structure criteria as theoretical basis to identify output-only modal parameters.CP-BSS algorithm is introduced into the field of modal identification,and combining Hilbert and support vector machine theory in statistical machine learning,the structural parameters and damage are explored.The main work is as follows:Firstly,the engineering background is introduced,and the research status and existing problems of modal parameter identification and damage identification methods are systematically studied.Then,the development of blind source separation theory and its application in structural modal identification are introduced.The basic principles of blind source separation and CP-BSS method theory are studied,including pretreatment method,mathematical model and separation process of mixed signals.Simulation results show that it has obvious advantages over traditional Independent Component Analysis algorithm.Secondly,CP-BSS is introduced into the structural modal parameter identification framework for the first time to construct mathematical equivalence.The feasibility of applying CP-BSS to structural vibration signals is studied,and the steps of applying CP-BSS to structural modal parameter identification are introduced.The validity of CPBSS applied to structural modal parameter identification is verified through a threelayer frame and a simplified offshore wind turbine foundation.The identification results of CP-BSS and Fast ICA algorithms are compared and analyzed to verify the adaptability of CP-BSS in damping and noise and the accuracy of identifying highorder modes.Thirdly,support vector machine theory and grey wolf optimization algorithm are studied and improved.In this paper,the theoretical model of support vector machine in machine learning algorithm is deeply studied,and a differential evolution improved grey wolf optimization algorithm is proposed to optimize the support vector machine model.Simulation experiments show that using differential evolution to improve grey wolf optimization algorithm can not only improve convergence and search speed in theory,but also enhance local search ability and diversity of solutions to avoid falling into local extremum.Finally,for the the offshore platform structure,a damage identification method based on CP-BSS and improved SVM is proposed.The basic principle of principal component analysis is studied,and the steps of applying CP-BSS to the damage identification of offshore platform are given.The support vector machine theory in chapter 4 is applied to the damage identification of offshore platform model,which verifies the reliability of the algorithm to identify structural damage and can accurately identify the damage location and damage extent of the structure.