Machine Intelligence for the Optimization of Value Networks Digitalization is changing the way companies collaborate with partners, suppliers and customers. There is an increasing number of global partners with complex and constantly changing relationships. These relationships can increasingly be mapped with digital data. The analysis and control of such value networks, consisting of companies, partners, suppliers and customers, is a challenge for the industry. In this project, value networks are to be mapped and analyzed digitally. The basis for this is making data available and processing it in the company as well as the further development of artificial intelligence processes. This includes the appropriate mapping of the value network and available data sources. Pilot projects with companies are aimed at better assessing and forecasting sales, the needs of customers and partners, and delivery times, and thus optimizing value creation relationships.

The challenge is the increasing complexity of value networks Companies are struggling with increasing complexity in their value networks. This is due to the increasing number of global partners and their extensive interdependencies, as well as a larger number of products with shorter life cycles. At the same time, the uncertainty factors of the global markets are increasing. This is reflected in challenges such as forecasting sales, requirements in the value network or delivery times and resulting effects such as the bullwhip effect or parts tourism. At the same time, digitized processes in networked IT systems generate large volumes of data that form the basis for analysis and optimization using artificial intelligence (AI) methods. Structured preparation of available data in the company is required, as is the preparation and further development of AI processes. The successful use of artificial intelligence for the optimization of value networks can be divided into three fields of action. HF1 Specification of cause-effect relationships: In the context of SCM and logistics, the logistical cause-effect relationships in a value network must be specified comprehensively. HF2 Specification and expansion of the IT infrastructure: In the context of digitization and Industry 4.0, IT systems must be specified. As a digital backbone, they represent the link between the physical world and virtual mapping. This is essential for understanding the existing data basis and its extension. HF3 Selection and development of AI methods: Simulation, AI, and operations research yield potentials for the analysis and optimization of value networks. For this purpose, computable models must be established and methods must be further developed.

Specification technique for value networks for analysis and optimization with AI methods The primary goal of the project is the analysis of typical effects in value networks and the automated analysis and optimization with AI methods. This is done on the basis of a domain-spanning specification of interdependencies in value networks. On the one hand, the description concerns the value network in the form of its elements, the relationships among each other as well as relationships to external elements. On the other hand, the computational representation of the value network is also described in terms of underlying IT systems, data sources, their properties and relationships to each other. In their entirety, the two partial models enable a description of the value network in the sense of a digital twin. The specification technique includes both semiformal models for simple cross-domain description and rapid incorporation of expert knowledge, as well as formal models as the basis for analysis with machine learning methods or computation within simulations. This enables the consideration of expert knowledge as well as knowledge derived from data. The specification technique is the basis for the further development of procedures for the automated analysis and optimization of value networks. This includes data-driven procedures from AI, machine learning and simulation, whereby the integration of experts and domain knowledge into data-driven procedures is necessary for success. The goal for pilot partners within the project is to solve problems within the value network by using AI procedures. This is essentially made possible by increasing the transparency of processes and effects, which is the basis for forecasting demand, sales and delivery times.

The diagram clearly shows that the project is implementing the necessary sub-steps for the implementation of fully autonomous systems. This is a necessary step on the way to the use of intelligent technical systems. The goal is solutions in the supply chain based on the use of AI processes. In this way, the already numerous existing data are to be refined. Furthermore, companies are to be enabled to use AI processes for the analysis and optimization of value creation processes. In this context, AI is understood as a tool to solve complex, data-based planning tasks in logistics. Here, the project makes use of existing ML methods and Big Data approaches, which are adapted to the application domain and/or developed further in a targeted manner. The overriding goal is not the integration of intelligence as an end in itself, but the solution of real challenges in practice (e.g., increasing the transparency of operations and effects in the value creation process). Overall, the project makes a significant contribution to expanding the leading position of the it's OWL cluster in the field of intelligent technical systems and to extending its innovation leadership.