Predictive Control With Active Disturbance Rejection for Elastic Drive Systems

Abstract

In this paper a model predictive disturbance compensation control concept is presented for an industrial combustion engine test bed. The provided controller represents an extension to an already existing predictive feedback controller and is utilized to improve control performance regarding shaft torque tracking and zero torque control. Close-to-reality load tracking, as it is desired for effective combustion engine development, requires both a stiff shaft connection between engine and dynamometer and high closed-loop control bandwidth. However, operation of the test bed close to the main rotor resonance follows which is accompanied by strong rotor excitations caused by the engine combustion torque. The combustion disturbance torque, which also acts as the main input to the predictive disturbance compensator, is obtained in this work from two different torque estimation approaches based on Kalman filtering and an appropriate FIFO memory. Additionally, active constraints for maximum dynamometer torque are utilized to further improve control performance and also to guarantee closed-loop stability. The proposed control scheme is tested in simulation and also in real time on an industrial test bed. Representative results are finally presented which verify the high efficiency of the developed predictive disturbance compensator.