The Applied Research Based On QPSO In Bridge Health Monitoring

Quantum particle swarm optimization (QPSO), as an artificial intelligence algorithm of global convergence, is becoming an emerging hot topic of structural health monitoring based on the intelligent algorithms in recent years, because of its advantages as follows: strong ability of the system identification, a few parameters need to be controlled, solving complex optimization problems and so on. In this paper, according to the features ofQPSO, based on the disadvantages of existing methods of modal parameters and damage identification, the paper presents new methods to identify structural modal parameters and damage through combining QPSO with Continuous Wavelet Transform (CWT) and Generalized Flexibility Matrix (GFM) and Time History Response (THR) respectively, and transferring the identification issue into optimization issue, and simplifying the identification process and improving the identification accuracy. Detail researches and results are as follows:(1) It is reviewed systematically that the theoretical significance, application background, research status, and shortcomings of existing methods about structural modal parameters identification and damage detection. Then the thoughts and main contents of bridge structural health monitoring based on QPSO are put forward.(2) The background, basic principles, calculation process, and optimization features of QPSO are elaborated. And the basic theory, main characteristics and manifestations of CWT and GFM are discussed briefly.(3) Taking into account the flaws of model parameter identification based on wavelet transform and Particle Swarm Optimization (PSO) under ambient excitation modal parameter respectively, a new method named QPSO+CWT is presented. Through CWT of the structural responses, multiple degrees of freedom (MDOF) systemic responses are changed into several single degree of freedom (SDOF) responses through CWT of structural output. Then SDOF responses are optimized by QPSO, and the modal parameters are identified simultaneously. On the one hand, the method is able to reduce the numbers of identified parameters and relax the constraints of each particle, on the other hand, it could optimize all the model parameters(frequencies, damping and mode shape) one time to simplify the calculation process and improve recognition accuracy. The numerical simulations of different structures are adopted to verify the effectiveness and viability of the method herein.(4) A new approach named QPSO+GFM is proposed based on the disadvantage of damage detection based on the sensitivity of flexibility matrix (FM) and generalized flexibility matrix (GFM). According to the relationship between structural generalized flexibility matrix difference and the changes of structural physical parameters before and after damage, QPSO is used to optimize and identify objective function constructed by GFM. And then the structural damage condition is identified accurately with the damage location and damage extent at the same time. The analysis results of several numerical simulations of different structures illustrate that the method is superior and more effective than the sensitivity of FM and GFM.(5) In terms of the shortcomings of damage identification based on PSO and the time history response (THR), the dissertation presents a new damage identification named QPSO+THR, gaining the damage location and extent simultaneously. The numerical simulations of different structures are applied to demonstrate the validity and excellent noise immunity of this method herein.