App settings:

Diebold Nixdorf

AI based Service Engineering

The optimization of various processes plays an increasingly important role for companies. Automation is constantly advancing and also requires continuously improved automated analysis processes. ATMs are highly complex mechatronic systems whose behavior is monitored by a number of actuators and sensor data. Up to now, patterns in the machine data have been mapped to possible defects by simple rules, without giving differentiated repair instructions to the service technician and without verifying the service instructions by results of the technician’s intervention. This is now to be guaranteed by Machine Learning (ML).

The aim of the project is to investigate use cases and associated algorithms on existing machine data and service platforms to implement an enrichment of so-called service calls or events. The intended work consists of developing interfaces to the AI marketplace platform (1), ensuring the validity and significance of the database (2) and designing AI applications for product development (3).

  1. A database is required for the implementation of ML algorithms. Thus, using various procedures, an AI can be taught to detect errors, classify error causes in a targeted manner and finally provide the service technician with precise repair instructions. This ensures that costs, time and spare parts can be used in a more targeted manner.

  2. In order to ensure that the AI classifies precise causes of failure, it will be trained and validated with non-identical data. When the AI has been trained and is able to make predictions, it will be tested on additional data sets.

  3. The design of AI applications for product creation will initially be developed within the framework of pilot projects for the AI Marketplace. On the research side, Fraunhofer IOSB-INA supports projects in the realization of AI applications and ML processes for different use cases.