Model Based System Identification for Hydraulic Deep Drawing Presses

Abstract

The paper describes the development of an automated system identification algorithm for the die cushion drive in hydraulic deep drawing presses. The algorithm could successfully be implemented on the drive controller to automatically identify the system parameters of the drive. Main aspect of the paper is the application driven development of an appropriate system model and identification algorithm with its implementation on the drive controller. It could be verified with experiments on a 2500 kN hydraulic deep drawing press. This thorough knowledge of the system model with its parameters shows a high potential to be further evaluated for system diagnosis and could also be used for system simulation and controller design. Key points are the limited processing power of the drive controller and the occurring signal noise. A grey-box system model was chosen and its parameters were identified by means of a recursive least square algorithm. The implementation on the drive controller required adaptions due to restricted cycle time and additional signal processing to reduce noise that will also be discussed in the paper.