machine learning

The logistics industry offers numerous opportunities for data-driven solutions, such as improving the part feeding problem in assembly line industries. A data-based approach for will lead to an improvement of cost-effectiveness through optimized processes, resource utilization, and consistent supply to the assembly line. The generated approach is a mixed integer programming model which considers limited storage space, uses constraints, and various cost factors related to transport, replenishment, and picking.

Artificial Intelligence (AI) can make a major contribution to the future viability of our economy and society—whether by improving existing processes or new products and services that promise greater efficiency, more robust structures and more climate protection. At present, however, SMEs in particular are still reluctant to use AI systems. The frequently cited reason is that data protection hurdles appear to be too high. This article discusses the opportunities of data-based value creation. The central question is how AI applications in industry can generate economic added value from data while maintaining data protection and security.

Energy efficiency is becoming increasingly important in all sectors of the manufacturing industry. Companies are currently feeling the pressure of exorbitant energy prices very clearly, as well as the additional challenge of becoming CO2-neutral by 2045. With technologies from the field of machine learning (ML), innovative solutions can be developed that enable energy-efficient product manufacturing. In this way, ML-supported process control can make a decisive contribution to increasing the sustainability and competitiveness of a company. Decisive for ML-supported process control are the process- and raw material- dependent parameters, which are significantly responsible for the quality of the final product. The subject of this paper is a procedure for analyzing the complex relationships between the relevant influencing parameters for increasing energy efficiency in the manufacturing industry. (Only in German)

Optimisation is an essential task in many situations. The class of evolutionary algorithms is a population-based, heuristic technique for optimisation. They allow the optimisation of multi-modal problems even with distorted search spaces. They can propose several solutions instead of just one. An important aspect of evolutionary algorithms is encoding search space candidates. In the optimisation of processes, this is a non-trivial task. This article describes a successfully tested encoding.

In factory operations, measuring equipment is often used without automatic storage or further processing possibilities of the measured value. In this case, employees must capture and process the measured values manually. In this article, an approach for the optical detection and digitization of measured values with the help of machine learning methods is presented. This aims to reduce the workload of the employees, avoid reading errors and enable automated documentation.

Supported by an intelligent assistance system, the sustainability and digitalisation of the tideindependent Industriehafen Bremen can be increased. To guarantee port security, a constant level of the isolated harbour is essential. There is great potential for increasing energy efficiency if the lock naturally waters the harbour basin at times of high tide instead of employing a pump station. This contribution shows how artificial intelligence and innovative software systems were used to develop an assistance system to improve existing procedures without making extensive changes to the existing port infrastructure. (Only in German)