Search

Abstract: In this research, state and imbalance fault estimation of a simple rotor-bearing system using Kalman filtering techniques has been investigated. The motion of a simple rotor- bearing system can be described by a set of coupled partial differential equations (PDEs); approximative equations of the motion PDEs are derived by variational formulation, where the PDE system is spatially discretized into a high dimensional ordinary differential equation (ODE) form. Optimization-based sensor selection algorithm for Kalman filtering is then applied to optimally choose among the large number of ODE model states to measure, such that specific requirements for state estimation performance are satisfied with a small number of sensors. For practical applications such as the rotor-bearing systems, fault estimation is usually one of the goals of system monitoring; augmented-state Kalman filter (ASKF) is preferred for its simple formulation, but at the cost of more intensive computation and greater numeric errors due to higher system order. Alternatively, optimal two-stage Kalman filter (OTSKF) provides an equivalent form of ASKF under certain algebraic constraint but with generally lower computation complexity and many practical advantages. Adaptive two-stage Kalman filter (ATSKF) is thus applied in this research for simultaneous state and fault estimation of the rotor-bearing system, and the optimal adaptive fading factor for OTSKF is designed using the innovation sequence which is equivalent to that of ASKF. Simulation results have demonstrated the effectiveness of ATSKF handling sudden imbalance fault occurrence during the operation of the rotor-bearing system using the optimally selected sensors.