The project goal is the development of analytical state estimators for the inline monitoring and fault diagnosis of servo presses. The project aims to develop analytical models of different servo press drive systems. In future, this should make it possible to digitally map the current condition of the press on the machine controller. If symptoms of wear occur within the drive system during the service life of the servo press, the machine operator will be informed directly about the position of the wear and an estimation of the remaining service life. This should enable maintenance work to be condition-orientated in the future and reduce production downtimes.

Motivation

Servo presses are among the most widely used forming machines in German industry and are used for a variety of forming processes. The use of servomotors is characterised by the availability of a large amount of high-resolution data that depicts a wide range of physical variables such as torques, motor currents, positions, speeds and accelerations. This measurement data are currently utilised for position control purposes, whereby their information potential far exceeds the limits of the controller. The potential for fault detection and fault diagnosis in servo presses has not yet been utilised. The development of analytical state estimators, which in future will utilise the existing measurement data on the controller, is intended to estimate the current machine state during operation of the servo press. This information should make it possible to plan maintenance and servicing work in a targeted manner as part of production planning and replace preventive maintenance of machine elements with condition-based maintenance. Costs for the maintenance of components subject to wear should therefore only be required if the condition of the components requires it. Furthermore, condition monitoring and condition-based maintenance should prevent unexpected damage and cost-intensive production downtimes.

Approach

The approach chosen in the research project for model-based condition monitoring using analytical state estimators is accompanied by better traceability of causal cause-and-effect relationships and has the potential to significantly increase the information content obtained from sensor data. This project therefore utilises white-box approaches, considering the static and dynamic effects in the drive systems of the servo presses. In this research project, a key role in model-based condition monitoring is played by the state estimators known from control engineering. They reconstruct or estimate the state of a real process or system using an analytical model and the known real input and output data.

The Luenberger observer and the Kalman filter, first introduced in 1960, are two of the best-known observers. Both are structurally parallel to the real system in their application and use the existing measurement data. The modelled system and the recursive mathematical algorithm of the estimator are used to calculate measurable and non-measurable states of the system. One of these states can be, for example, the friction within a bearing or the ram guidance. If the friction increases due to wear in the machine elements, this can be recognised by the state estimator and thus used for fault detection and diagnosis. This information can then be made directly available to the machine operators. By recognising deviations from the normal condition at an early stage, maintenance work can be planned.

Acknowledgement

The research work presented here is part of the IGF project no. 01F23257N of the research association Europäische Forschungsgesellschaft für Blechverarbeitung e.V. (EFB). This is funded by the German Federal Ministry of Economics and Climate Protection (BMWK) via the German Aerospace Center (DLR) as part of the programme for the promotion of joint industrial research (IGF) on the basis of a resolution of the German Bundestag.

We would also like to thank all the industrial partners who support the research project in the project support committee.

Funded by

Project Partner

• Andritz Kaiser GmbH

• Bosch Rexroth

• Haulick + Roos GmbH

• H&T ProduktionsTechnologie GmbH

• LASCO Umformtechnik GmbH

• Nidec SYS GmbH

• Schuler Pressen GmbH

• Strack Norma GmbH & Co. KG

• Siemens AG

• Synchropress GmbH

• Hans Berg GmbH & Co. KG

• KODA Stanz- und Biegetechnik GmbH

• Röcher GmbH & Co. KG

• Profimetall Engineering GmbH

• Mercedes Benz