System Identification For Linear And Nonlinear Structure Based On Revised Multi-particle Swarm Co-evolution Optimization Algorithm

Structural damage detection(SDD)is a critical issue of structural health monitoring(SHM)and it can often be treated as a system identification(SI)problem.A number of SI methods have been developed,and the vibration-based SI methods have attracted considerable attention for two decades.A large number of vibration-based methods have been studied for a long time and have shown efficiency,and the majority of the works focuses on estimating physical parameters using the measurement of both the input excitation and the structural response for a small structural system.However,the time history of dynamic input forces,for example,wind and traffic,are often difficult to be directly measured;it is also not easy to obtain complete structural response(either accelerations,or velocities or displacements).Moreover,with the larger of civil engineering structures,it will inevitably increase the difficulty of identification.In addition,most civil engineering structures exhibit different levels of nonlinearity when subjected to long-term dynamic loads or catastrophic events,such as earthquake and typhoon.Generally,the structures will be in nonlinear state with the appearance of structural damage.The existing SI methods are mainly dependent on the linear structural theory,however,they can not detect the structural state(linear or nonlinear)and identify the systems of large-scale linear and nonlinear structures.Therefore,the new SI method is investigated by using the intelligent optimization algorithm and signal analysis technology in this thesis.The primary studies and achievements are as follows:(1)This thesis studies on structural state detection strategy based on the only structural output measurements,and proposes a new detection method.A numerical example and a laboratory test are conducted to verify the effectiveness of the proposed method.The results show that the proposed method can not only detect the presence of the nonlinearity,but also quantify its degree.(2)Due to the fact that the multi-particle swarm co-evolution optimization(MPSCO)is likely to end up with a premature convergence,an improved work is carried out in this thesis through adding a novel processing mechanism of worst particle in the iterative procedure of MPSCO.The efficiency and applicability of the revised multi-particle swarm co-evolution optimization(RMPSCO)is validated by linear and nonlinear numerical simulations and experimental study.The results show that,compared with other optimization algorithms,RMPSCO shows a better stability and a higher accuracy of parameter identification for both linear and nonlinear structural systems.(3)The system identification methods for small-scale and large-scale linear structures with only incomplete sets of output measurements are studied in this thesis.The efficiency and applicability of the proposed technique are validated by two numerical examples and a shear-type frame test under different conditions of data availabilities and noise corruption levels.The results show that the proposed technique is powerful,robust and efficient in the simultaneous identification of the structural parameters and excitation even using an incomplete set of noise-contaminated structural response measurements.(4)This thesis studies on the SI strategy for large-scale linear structures with limited output measurements and known excitation,and proposes a three-stage SI method.A numerical example and a laboratory test are conducted to verify the effectiveness of the proposed method.The results show that the novel three-stage method is able to effectively and quickly identify the damage instant,location and extent as well as the relevant physical parameters,furthermore,the proposed method has excellent noise tolerance and robustness.(5)To solve the problem of low SI accuracy of nonlinear structures,this thesis presents a new two-stage nonlinear structural SI method based on RMPSCO and fractal theory for frame structures under earthquake excitation.More specifically,in the first stage,the fractal theory is used to roughly locate the structural nonlinearity;in the second stage,RMPSCO is used to precisely locate the structural nonlinearity and identify the related model parameters.Moreover,the method can be extended to SI of large-scale frame structures when it is integrated with substructure technique.Finally,two numerical examples with Bouc-Wen model and a laboratory test are presented to verify the effectiveness of the proposed method.The results show that the SI of nonlinear frame structures,involving nonlinear location and quantification,can be implemented successfully,and meanwhile the nonlinear features of the structure can be identified accurately.In summary,this thesis presents effective SI methods for small-scale and large-scale structures with incomplete measurements in the field of real-time structural health monitoring.Furthermore,it provides a valuable reference and technique support for scholars and readers interested in SI of nonlinear structures and the nonlinear seismic mitigation and isolation structures in the future.