Adaptive Workflow Management for Cyber-Physical Systems - A Hybrid Approach Combining Experience-Based Learning and Automated Planning for Industry 4.0. Malburg, L. Ph.D. Thesis, Universität Trier, 2025.
abstract   bibtex   
The manufacturing industry is confronted with challenges caused by shorter innovation and product lifecycles, increasing complexity of production due to individualization and customization, and production disturbances of various kinds. To address these challenges, more autonomous and intelligent production systems are required, enabling the manufacturing industry to react more flexibly to these changing conditions. One goal of the Fourth Industrial Revolution, also known as Industry 4.0 (I4.0), is to enhance the digitalization and connectivity of high-level information systems with low-level production processes. This connectivity is enabled by the fact that an increasing number of devices and machines are connected to the Internet, known as the Internet of Things (IoT). According to this, IoT is a key technology for realizing Cyber-Physical Systems (CPSs), linking the physical world, i.e., devices and machines, with the digital world. In the manufacturing industry, Cyber-Physical Production Systems (CPPSs) as a special kind of CPSs are applied to optimize production processes and to increase efficiency and flexibility in manufacturing. Although CPPSs are already used in industry and there is a growing amount of research in this field, a bidirectional coupling between high-level information systems and low-level production processes is often still missing. Consequently, the IoT sensor data generated during production cannot be fully utilized to monitor low-level production processes, which in turn makes it more difficult for high-level information systems to detect emerging situations at an early stage and to perform real-time adaptations. The approaches for enabling flexibility and adaptability of processes in CPPSs are still in their infancy and other available approaches do not adequately consider the specific characteristics of CPPSs such as unanticipated behavior or physical errors. Consequently, current production processes are often still too rigid, which makes it difficult to react to changing environmental conditions or disturbances in real-time and in an automated way. This thesis aims to contribute to the aforementioned problems by developing an approach for adaptive workflow management for CPPSs by using Artificial Intelligence (AI) techniques. For this purpose, a systems architecture is developed that enables the bidirectional coupling between high-level information systems and low-level processes. Based on this coupling, Business Process Management (BPM) methods are used to model production processes on an abstract level and in a more structured way, allowing the flexible composition and execution of low-level production processes. Moreover, methods from BPM enable the utilization of IoT sensor data for process monitoring and analysis, building the basis for detecting emerging situations and disturbances in production processes earlier. The presented adaptive workflow management approach enables resolving such situations by ad hoc adaptations of production processes in near real-time. This is achieved by combining search-intensive AI planning and knowledge-intensive Case-Based Reasoning (CBR). One main advantage of this combined approach is that already experienced problem situations in the form of performed ad hoc adaptations (called cases) can be reused to solve new upcoming similar problems. In addition, the search-intensive AI planning solves problems that have not yet been experienced as cases. In this way, the combined approach helps to limit both the high knowledge acquisition and modeling efforts and the high computational complexity compared to using search-intensive AI planning solely. At the same time, it also increases problem-solving competence in situations where CBR alone cannot yet solve these problems. Finally, the approach enables incremental self-learning and, thus, improves the problem-solving competence continuously. In addition, the use of CBR enables the integration of domain experts and their expertise into the combined approach. The experimental evaluations indicate the suitability of the proposed approach for performing ad hoc adaptations in CPPSs. For this purpose, a learning factory has been used as a test bed to simulate production processes of a real manufacturing environment. Using such learning factories, the characteristics of real-world production environments are considered, and at the same time, such learning factories allow a more cost-effective and easier realization of I4.0 research. For evaluation, the production processes have been modeled and executed in the learning factory. In addition, failure situations have been generated to test whether these failures can be detected with enhanced monitoring techniques and to adapt production processes using the proposed adaptive workflow management approach.
@phdthesis{Malburg_PhDThesis_2025,
    abstract = {The manufacturing industry is confronted with challenges caused by shorter innovation and product lifecycles, increasing complexity of production due to individualization and customization, and production disturbances of various kinds. 
To address these challenges, more autonomous and intelligent production systems are required, enabling the manufacturing industry to react more flexibly to these changing conditions. 
One goal of the Fourth Industrial Revolution, also known as Industry 4.0 (I4.0), is to enhance the digitalization and connectivity of high-level information systems with low-level production processes.
This connectivity is enabled by the fact that an increasing number of devices and machines are connected to the Internet, known as the Internet of Things (IoT).
According to this, IoT is a key technology for realizing Cyber-Physical Systems (CPSs), linking the physical world, i.e., devices and machines, with the digital world.
In the manufacturing industry, Cyber-Physical Production Systems (CPPSs) as a special kind of CPSs are applied to optimize production processes and to increase efficiency and flexibility in manufacturing. 
Although CPPSs are already used in industry and there is a growing amount of research in this field, a bidirectional coupling between high-level information systems and low-level production processes is often still missing.
Consequently, the IoT sensor data generated during production cannot be fully utilized to monitor low-level production processes, which in turn makes it more difficult for high-level information systems to detect emerging situations at an early stage and to perform real-time adaptations.
The approaches for enabling flexibility and adaptability of processes in CPPSs are still in their infancy and other available approaches do not adequately consider the specific characteristics of CPPSs such as unanticipated behavior or physical errors.
Consequently, current production processes are often still too rigid, which makes it difficult to react to changing environmental conditions or disturbances in real-time and in an automated way.

This thesis aims to contribute to the aforementioned problems by developing an approach for adaptive workflow management for CPPSs by using Artificial Intelligence (AI) techniques. 
For this purpose, a systems architecture is developed that enables the bidirectional coupling between high-level information systems and low-level processes.
Based on this coupling, Business Process Management (BPM) methods are used to model production processes on an abstract level and in a more structured way, allowing the flexible composition and execution of low-level production processes.
Moreover, methods from BPM enable the utilization of IoT sensor data for process monitoring and analysis, building the basis for detecting emerging situations and disturbances in production processes earlier.
The presented adaptive workflow management approach enables resolving such situations by ad hoc adaptations of production processes in near real-time.
This is achieved by combining search-intensive AI planning and knowledge-intensive Case-Based Reasoning (CBR).
One main advantage of this combined approach is that already experienced problem situations in the form of performed ad hoc adaptations (called cases) can be reused to solve new upcoming similar problems. 
In addition, the search-intensive AI planning solves problems that have not yet been experienced as cases.
In this way, the combined approach helps to limit both the high knowledge acquisition and modeling efforts and the high computational complexity compared to using search-intensive AI planning solely.
At the same time, it also increases problem-solving competence in situations where CBR alone cannot yet solve these problems.
Finally, the approach enables incremental self-learning and, thus, improves the problem-solving competence continuously.
In addition, the use of CBR enables the integration of domain experts and their expertise into the combined approach.

The experimental evaluations indicate the suitability of the proposed approach for performing ad hoc adaptations in CPPSs.
For this purpose, a learning factory has been used as a test bed to simulate production processes of a real manufacturing environment.
Using such learning factories, the characteristics of real-world production environments are considered, and at the same time, such learning factories allow a more cost-effective and easier realization of I4.0 research.
For evaluation, the production processes have been modeled and executed in the learning factory.
In addition, failure situations have been generated to test whether these failures can be detected with enhanced monitoring techniques and to adapt production processes using the proposed adaptive workflow management approach.},
    year = {2025},
    title = {{Adaptive Workflow Management for Cyber-Physical Systems - A Hybrid Approach Combining Experience-Based Learning and Automated Planning for Industry 4.0}},
    school = {Universität Trier},
    author = {Lukas Malburg},
    keywords = {Adaptive Workflow Management, Cyber-Physical Production Systems, Internet of Things, Case-Based Reasoning, Automated Planning}
}
Downloads: 0
About
Documentation
Keywords
Search
Users
Terms and Conditions of Use
Privacy Policy
Sitemap
© BibBase.org 2021