Process optimization is crucial in industries with high energy demands like industrial material processing. Neural networks are commonly used for this purpose, but their performance deteriorates over time due to material wear and changes in data distribution. This study addresses these challenges by enhancing modeling, prediction, and optimization through continual learning techniques that detect and adapt to concept drift. This approach maintains model performance despite changing conditions and represents the first contribution to adaptive optimization in industrial material processing. The methodology’s effectiveness is validated with real-world data.

The increasing virtualization of automation software and its decoupling from hardware introduces a significant change in industrial automation. This transition allows classical operational technology software instances, such as programmable logic controllers, to coexist on the same host system alongside applications from various domains. Such coexistence offers substantial potential for a more efficient interconnection between applications from different areas, overcoming typical limitations associated with conventional interfaces. However, current middleware solutions for automation applications are often proprietary and not standardized. In this paper, we present an initial concept for an intra-host middleware based on the OPC Unified Architecture, employing a centralized data directory combined with a shared-memory-based transport mechanism. The proposed concept aims to enable efficient, low-latency, and scalable communication, thereby facilitating seamless interaction between diverse automation applications and promoting the use of microservice architectures.

To meet the dynamic demands of Industry 4.0, manufacturing systems require enhanced adaptability and interoperability, particularly in light of the transition from Machinery Directive 2006/42/EC to Machinery Regulation 2023/1230/EU. This transition emphasizes the need for machine-readable and digital documentation standards. However, integrating these standards with capability modeling frameworks such as the Capability-Skill-Service model (CSS model), the Open Platform Communications Unified Architecture (OPC UA) and the Asset Administration Shell (AAS) remains challenging, which limits seamless interoperability and dynamic reconfiguration. This paper addresses these issues by presenting a methodology to automate capability generation using large language models. This approach standardizes technical documentation and bridges gaps through automation. Aligning the CSS model with the OPC UA and AAS frameworks enables integration into Manufacturing-as-a-Service platforms, driving innovation and efficiency in modern manufacturing ecosystems.

The increasing dynamics and complexity of today’s manufacturing environments require greater flexibility and resilience to meet the demands of producing multi-variant products, including batch size one, with minimal additional effort. The Capability, Service and Skill Model provides an answer to this problem by decoupling processes and the resources used. However, especially small and medium-sized enterprises face the challenge of systematically modeling their machines and systems in terms of capabilities and skills, as the required knowledge is often not available in a structured form and is often tied to individual key personnel. To ensure a consistent and practical description of capabilities, a method based on historical data is proposed to help companies integrate their manufacturing resources into a networked system of manufacturing service providers. On this basis, historical manufacturing features are systematically examined to evaluate the potential to use them to determine capabilities.

This paper is dedicated to solving the problem of concept drift in industrial plants using artificial intelligence methods. For this purpose, methodological approaches and procedures are considered and analyzed. Based on the findings, reference architectures were developed at different abstraction levels that can be used in an industrial environment and enable continuous machine learning. Continuous machine learning offers the possibility of adapting to dynamic changes in the production environment, which are reflected in constantly changing data sets. Through a combination of machine learning techniques, a novel and practical framework for continuous learning, also known as lifelong learning, is presented. The integration of problem-focused machine learning methods is advancing in production, e.g., predictive maintenance, process optimization or fault detection. Thereby, fully or semi-automated adaptations to changing environments requiring continuous improvements are less often explored, although practical use cases often require adaptive capabilities as the physical data distribution may change over time. In this paper, the application was continuously improved based on case studies and empirical results, and finally validated with a quality assurance application. Various methods and approaches for detecting concept and data deviations, retraining, packaging and model updating had to be investigated, which led to the question of what a real industry-oriented implementation could look like. The result is a reference architecture that can run on cloud and edge computing resources. This reference architecture is validated in real-world application in the parquet production sector, proving its feasibility and efficiency.