Physical Parameter Of System Identification Method Based On Subspace Identification Method

In recent years, more and more structural health monitoring systems have been installed to bridges, tunnels, large span spatial structures, high-rise structures, offshore platforms and other large civil engineering structures. Structural health monitoring systems can monitor and early warn for the operational status of the engineering structures, and they play important roles in improving operational efficiency of the engineering structures and protecting people's lives and property. However, parameter identification of system, as one of the core technologies of structural health monitoring system, has not been solved fundamentally. And it seriously limits the development of structural health monitoring system. Parameter identification of system is divided into physical parameter identification and modal parameter identification. Physical parameter of system is the direct representation of the current state of the system and a parameter that needs to be applied directly to the system reliability evaluation. Modal parameter of system, a function of the physical parameter, can indirectly reflect the changes in the physical condition of the system. So it is able to determine qualitatively and quantitatively the changes in the state of the system. Physical parameter and modal parameter are both significant contents of structural health monitoring.The paper mainly studies the theories and the application of subspace identification method. Both theoretical analysis and numerical experiments demonstrate the validity of the modal parameter of system, identificated by subspace identification method. Moreover, combined with Harmony Search algorithm, which is a novel intelligent optimization algorithm, a method to identify the physical parameter of system is proposed. Thus it can be used to determine qualitatively and quantitatively the changes in the state of system, achieving the purpose of health monitoring. The main contents are as follows:1) In the traditional method of modal parameter identification, it needs excitation equipment to excite the system to obtain the input and the output of the system. In this way, modal parameter identification of system can be achieved. Therefore, the paper introduces a method called stochastic subspace identification method, based on environmental incentive. A numerical simulation example is given to show that when the input data is unknown, the method can distinguish modal parameter of system according the output data.2) Studying thoroughly the basic theory and the derivation of stochastic subspace identification method, an improved data-driven stochastic identification method based on Householder transform is proposed. A numerical simulation example shows that the improved method can speed the calculation effectively, ensuring the accuracy.3) The impact of system order and dimension of the Hankel matrix on stochastic subspace identification method is investigated. On the basis of the existing main methods of system order determination, with a little modification, singular-values-difference slope method and slope normalization method of increment of the singularity entropy are proposed. In addition, the basic Harmony Search algorithm is creatively combined with determining the appropriate dimension of the Hankel matrix.4) Physical parameter of system identification method, which is combined stochastic subspace identification method with Harmony Search algorithm, is studied. In this method, the physical parameter identification of the system is considered as an optimization problem, defining the objective function constructed by modal parameter of system, which is identified by stochastic subspace identification method. A numerical simulation example of three-freedom structure result indicates that the proposed method can effectively identify the physical parameter of system. In the example, the maximum error of identification of system stiffness is 3.52%. Then, the objective function and the definition of parameters in Harmony Search algorithm are improved. At this time, the minimum error of identification of system stiffness is as low as 0.03%.5) Stochastic subspace identification method is applied to the field of damage identification of systems. A step-by-step method is proposed in this paper, based on the orthogonal property of subspace, prediction error analysis of Kalman filter and identification of the physical parameter of system with improved Harmony Search algorithm, to identify the damage of system.