Deduction of Goal-Oriented Minimum-Order Models for Advanced Motion Control on the Example of Large Industrial Drives

Adjustable speed drives are used in a wide variety of applications which in many cases include loads with relevant mechanical flexibility. Unwanted mechanical oscillations deteriorate the process quality and threaten to damage the drive train. Due to limited overload capabilities of mechanical components in high power drive systems, the reduction of mechanical strain is considered a crucial control objective. Therfore, acurate models reproducing all important system dynamics of the drive train are required, which, at the same time, have the lowest possible dynamic order to be suitable for online as well as offline optimization and model-based control. Within the scope of this paper, the influence of higher eigenmodes and the shortcomings of two- mass drive train models for controller design and performance evaluation is discussed based on simulative investigations on multi-megawatt rolling mill drives. A strategy for determining all performance-relevant eigenmodes of a drive train is presented.