An inverse-forward dynamics for real-time operational clamping force prediction of BLAC Motor

Abstract

This study presents an inverse-forward dynamics for clamping force prediction of brushless AC (BLAC) motor attached to motor driven power steering system (MDPS) in operational condition. The clamping force that is important to safety as well as NVH is difficult to experimentally measure during the operation, and electro-magnetic force that is a main loading condition of MDPS has also a lot of uncertainties in the system level. In this work, equivalent bearing force is employed as a replacement for the electromagnetic force, and it is indirectly predicted by using measurable acceleration data in the manner of inverse source identification. The clamping force between the BLAC motor and the supporting structure is then computed by the forward time integration algorithm. A new inverse-forward dynamic formulation combining implicit generalized-$\alpha$ time integrator and Tikhonov regularization is introduced to get the equivalent bearing forces and the clamping forces from noisy acceleration data. Model updating of the BLAC motor and the test bed is also necessary to accurately compute the force components through the proposed inverse-forward process. Among the various reduced-order modeling (ROM) techniques, Craig-Bampton component mode synthesis (CB-CMS) is employed. CB-CMS that has modal-nodal Degrees of Freedom (DOFs) in the reduced system matrices allows direct coupling between measured data and desired force data in the physical domain as well as efficient force computation. The accuracy and efficiency of the proposed process are evaluated through experiments with the BLAC motor test bed.