The implementation of data-driven methods from prescriptive analytics and machine learning in the automation projects of the industrial solution provider ABF GmbH is the core objective of the project. Together with RISC Software GmbH, the aim is to enable on-line optimization of decisions in the control processes in particular.
ABF’s systems control complex production environments and intralogistics systems based on the OneBase system platform. When new products, production facilities, transport equipment or changed operating modes require adjustments to software and regulations in existing projects, this places high demands on personnel both on the customer side and at ABF. After all, their effects on the currently running overall system are usually difficult to predict, especially in special cases.
OneBase supports such adaptations by means of simulation functions with which the production process can be recorded and repeated off-line for test purposes before changed versions are used in production. However, this involves considerable effort and the introduction of changes during operation is therefore always associated with stress, especially because most customers’ systems run around the clock.
Using current methods from the field of prescriptive analytics and machine learning, the aim is to improve and simplify these change processes – through automatic adaptation with a corresponding reduction in manual intervention in the logic and rules. This includes machine learning, simulation and optimization using current heuristic research approaches. These methods are then used to improve order planning and allocation. The quality of the solution is also to be increased for allocation and routing problems, where the time component is to be taken into account, by means of suitable forecasts of the future order situation.
This project is funded by the FFG.
Senior Mathematical Optimization Specialist
