THE EXTENDED KALMAN FILTER IN THE FREQUENCY DOMAIN FOR THE IDENTIFICATION OF MECHANICAL STRUCTURES EXCITED BY SINUSOIDAL MULTIPLE INPUTS

Abstract A modal parameter identification method applied to mechanical structures excited by correlated sinusoidal multiple inputs was developed. The algorithm is based on the same formulation of the extended Kalman filter, applied as a system parameter identifier in the frequency domain to mechanical structures subject to excitations characterised by an inherently high degree of correlation. The algorithm was validated by using simulated data on a multi-degree-of-freedom system. The tests demonstrate that the proposed technique is of practical application value. The method is devoted to the identification of the modal parameters of supporting structures of rotating machinery, using data obtained during the normal operation of the machines. Nonetheless, it can have an interest for more general applications in the field of the experimental modal analysis. This study was undertaken in the framework of the BRITE EURAM III project MODIAROT (MOdel based DIAgnosis of ROTors in power plants).