Hyyryläinen J.

The aim of the study was to develop and to demonstrate an on-line system identification procedure and related filters, which are needed for active control of vibrations in rotating machines. Adaptability in vibration control is needed to guarantee optimum performance if operating conditions change. If adaptive control is applied in rotor vibration damping, it is important to identify the rotordynamics on-line. The quality of the model is tightly linked to the quality of the measured data. Conditioning of the data is necessary. In this study, system identification relates to dynamic behavior between the damping actuator and rotor vibration. The study was initiated in the simulation environment. 3-D dynamical model of a rotor was imported as one block into the Simulink model. The vibration control method used in Matlab was HHC, Higher Harmonic Control. HHC is an open loop control method, where the sinusoidal control signal is synchronized according to the measured vibrations and the frequency response model of the system. The frequency response has to be modeled at the same time the first harmonic vibration component is damped. Because the dynamics of the rotor may be time variant, the frequency response has to be identified on-line. On-line identification tests were performed with the RLS (Recursive Least Squares) algorithm. Operating conditions were changed so that variation in dynamics would be emerged. The experimental part of the study was performed on a rotor test rig. Test rig consists of a rotor supported by journal bearings, an electrical drive to run the rotor, displacement transducers to measure the movements of the rotor, and a rigid base. The rotor kit was equipped with a magnetic actuator designed for active vibration control. On-line identification tests with the RLS algorithm were performed with and without active vibration control and rotation frequency was changed.