Time Domain Method Of Parameter Identification For Civil Engineering Structures

The theories on parameter identification for civil engineering structures serve as basis for the work of establishing structural health-monitoring system, identifying possible damages and evaluating structural conditions, thus their significance in maintaining security and normal performances for structures on service. This thesis mainly discusses the time-domain methods for identification of structural physical parameters, for the purpose that, by means of the measurement data of structural dynamic responses, people can trace characteristic physical parameters which can effectively reflect change of real conditions of those observed structures.In the field of physical parameter identification in time domain, some fundamental concepts and main research methods are summarized at first, which help establish the theoretical frame for the consequent research. Then based on an existing identification algorithm, an improved version of the algorithm is proposed. The improved algorithm is grounded on the finite element theory and the sensitivity analysis method, and it can identify finite elements'physical parameters without input information, achieving much higher identification efficiency. Methods are further proposed on how to employ the improved algorithm to identify characteristic parameters in substructures and in nonlinear parameter systems. Such theoretical extensions are quite meaningful for improving identification efficiency, lightening the workload of response measurement and data preparation before identification, and extending the scope of types of structures that can be identified by the algorithm.To illustrate the practical application of the identification algorithm in real projects and to evaluate its performances, four types of structures are identified in the thesis by numerical simulations. Also the treatment of damping in the finite element models and the influences of unknown inputs on the identification results are discussed. Some conclusions can be finally arrived, that the improved identification algorithm is applicable to a wide range of structural types, including large-scale structures, substructures and structures in plain stress state etc., and it can exactly identify physical parameters of finite elements without the input information if not considering noises buried in measurement data.Since a certain level of measurement noises is normally inevitable in the real world, it seems necessary to enable the identification algorithms to resist a certain...