Estimação de parâmetros de sistemas mecânicos no domínio do tempo utilizando-se o filtro de Kalman em conjunção com o estimulador das variáveis instrumentais.

Abstract

The objetive of this work is to investigate techniques of parameter estimation of mechanical systems in time domain, with enphasis to mechanical systems that can be eventually modeled by means of massHSpring-viscous damper ( dashpot) with a degree of freedom, submitted to small disturbances around the position of equilibrium. With this objective, simulations are processed in time domain to get state vector of the system ( data X and numerical methods for the solution of linear systems used herein parameter estimation process are commented In this work, important aspects about the search criterion for the signals sampling rate and its effects in the estimated coefficients are commented. Commentaries about the noise effects on output signal, its statistical properties and its influence on the estimated coefficients are commented. Theorical aspects about System Theory, Least Square Estimation, concepts of estimation methods ( Least-Square Method, Instrumental Variables Method X concepts of filtering using Kalman Filter and an introduction of spectral analysis of signal and systems can be added in this works Finally, the results of the application of parameters estimation technique are presented and compared with the theorical values attained from simulation. The efficiency of the Instrumentals Variables Method in conjunction with the Kalman Filter are discussed. Also emphasis has been given on the influence of excitation source used and its advantage with respect to other kinds of sources normally used.