Real-Time Nonlinear Dynamics Performances of Vertical Rotor System

Rotor imbalance in geotechnical centrifuges induces severe vibrations that degrade bearings, fatigue supporting structures, and may cause catastrophic failures. Although conventional balancing techniques are mature, centrifuge-specific challenges persist due to heterogeneous sensor modalities, measurement noise, and potential sensor faults. This study aims to develop a robust multi-sensor data fusion (MSDF) algorithm for real-time dynamic balancing of vertical rotor systems, addressing sensor-density and heterogeneity issues without hardware reconfiguration.

A full-order nonlinear dynamic model is derived for the vertical rotor system, and the unbalance response is analyzed for dynamic balancing applications. The proposed MSDF algorithm synergistically processes vibration signals and shaft-orbit trajectories to automatically extract effective information while suppressing redundant or corrupted channels. The framework accommodates multi-speed, single-plane, dual-plane, and general multi-plane balancing. Experimental validation is conducted on three rotor configurations.

A single balancing run reduces vibration by over 78%, with contaminated data contributing less than 4% to residual calculations. The proposed MSDF approach provides a robust, adaptive solution for intelligent maintenance of high-speed centrifuges.