ADIMA: Adaptive Assistive System for Maintenance of Intelligent Machines and Plants

Motivation

In an ever changing world, in- dustrial systems become more and more complex. Machine feedback in the form of alarms and notifications, due to its growing volume, becomes overwhelming for the operators. In addition, expectations in relation to system availability are growing as well. Often downtime of one machine results in a halt or delay of the whole value adding process causing high financial losses. Therefore, in order to maximize the availability, new maintenance concepts are necessary to prevent or minimize unplanned operation interruptions by introducingappropriatemaintenance measures or by reducing the repairing time in case of failure.

Objectives

The aim of the project was to develop an adaptive and intelligent assistance system for the main- tenance and troubleshooting of machines and plants. The system to be developed should be able to collect sensor data and alarm messages from industrial plants and analyse them using machine learning approaches to automatically generate maintenance guidelines. This includes recommendations for action which is appropriately visualised based on the available contextual information. The project combines the research areas of industrial communication and information modelling, Artificial Intelligence and human-technology interaction.

Research Activities 

In the final stage of the project, the prototype implemented in the SmartFactoryOWL on the basis of an industrial machine provided by the project partner Kannegiesser was finalised. The implemented asset administration shell (AAS) provides access to dynamic information, such as sensor data and messages from the machine, as well as static information, such as documentation and product descriptions. Using this information, the machine learning system (MLS) is able to learn a model of the causalities between the machine’s messages. The generated model enables the MLS to analyse newly occurring messages. The result of this analysis and further information from the AAS are then visualised via a mobile human-machine interface consisting of data glasses, tablet PCs, and smartwatches. Through the use of Augmented Reality, the information can be displayed spatially in the environment of the industrial plant. After completing the implementations, the overall system was presented to potential stakeholders in order to gather information on the applicability and acceptance of the applied concepts and technologies.