| In search for better procedures which identify an experimental modal model from measurement data, some new analysis methods, designated direct parameter model identification methods, are developed. These methods are designed to identify an experimental modal model by modelling the measurement data directly using the constitutive differential equations. The parameters in a fequency domain representation, or in a discrete time finite difference approximation, of the differential equations are estimated, using multiple input multiple output data simultaneously. A modal model is then obtained by uncoupling the estimated form of the differential equations. This enables the identification of experimental modal models with highly coupled and pseudo-repeated modes.; In applying the developed methods, frequency response functions, or impulse response functions, free decay data or forced response data can be used. The procedures developed for analyzing time domain data are shown to be numerically better conditioned and to execute faster than the corresponding procedures for frequency domain data. In the time domain, sampled force input and response sequences can also be analyzed directly, therefore avoiding any truncation errors introduced by a finite discrete transform between time and frequency domain.; Numerous simulated test cases are discussed to demonstrate the various characteristics of the developed direct parameter model identification methods, in particular the capability to obtain improved experimental modal models by simultaneous analysis of multiple input data and direct time domain processing. The applicability on experimental data is demonstrated using impulse response functions for several reference locations of a circular plate structure and an aircraft structure, both structures exhibiting pseudo-repeated modes. |
