Design Of Virtual Dynamic Analysis Instrument Based On Modal Parameter Identification

This dissertation intends to develop a set of virtual instrument of dynamic signal analysis which can run independently or perform as subprogram, based on modal parameter identification of vibrant system joining with virtual instrument design technique.The dissertation mainly study classical curve fitting of frequency response measurements using the rational fraction polynomial method and rapidly developing poly-reference implementation of the least-squares complex frequency-domain estimator. A modal analysis program is developed based on these methods. Stabilization chart is imported to select poles accurately adopting global curve fitting method. As to the poly-reference implementation of the least-squares complex frequency-domain estimator, right matrix-fraction description model is built to estimate the system poles. Then stabilization diagram is set up and analyzed to automatically determine natural frequencies, modal damping and modal participation factors. Finally mode shapes are identified based on the least square theory.In order to hunt for actual system poles in the built stabilization diagram, fuzzy clustering techniques is introduced. Some clusters are assorted according intrinsic properties of all poles. Then the actual system poles are calculated as gist of poles selecting. Normal clustering techniques need improvement to pick poles automatically and eliminate difference.