Modelling and identification for control of gas bearings

• The paper shows an identification based procedure for modelling of gas bearing systems.
• The procedure uses grey-box system identification techniques and can be applied in situ.
• A linear parameter varying model is derived from the identified models of a test rig.
• A controller is designed using the model and shown to enhance the damping significantly.
• All results are validated experimentally to confirm the suitability of the procedure.

Gas bearings are popular for their high speed capabilities, low friction and clean operation, but suffer from poor damping, which poses challenges for safe operation in presence of disturbances. Feedback control can achieve enhanced damping but requires low complexity models of the dominant dynamics over its entire operating range. Models from first principles are complex and sensitive to parameter uncertainty. This paper presents an experimental technique for “in situ” identification of a low complexity model of a rotor–bearing–actuator system and demonstrates identification over relevant ranges of rotational speed and gas injection pressure. This is obtained using parameter-varying linear models that are found to capture the dominant dynamics. The approach is shown to be easily applied and to suit subsequent control design. Based on the identified models, decentralised proportional control is designed and shown to obtain the required damping in theory and in a laboratory test rig.