During the first 18 months, the DMaaST project established the technical foundations for smart, resilient, and self-adaptive manufacturing networks. Activities focused on the specification, design, and initial implementation of core technologies:
1. During this reporting period, the project established the technical foundations of the DMaaST data and intelligence stack. Progress was made in the definition and implementation of the data layer, including decentralized knowledge graphs and ontologies enabling secure, trusted, and real-time data integration across organisational boundaries. These elements were built and/or selected for providing the interoperability required for the rest of the layers/modules within DMaaST. Work is in progress towards building the data streaming services for these components and the main technologies to be used habe been selected.
2. Initial frameworks were defined to connect manufacturing service–level and value-chain–level DTs. At production-line level, detailed models of machinery, processes, and logistics were developed, capturing operational constraints and interdependencies to support realistic simulation of shop-floor behaviour. In parallel, value-chain digital twins for the aeronautics and electronics sectors were specified, modelling inter-company interactions and disruption propagation. The architectural definition and interfaces between both levels were finalised.
3.Learning-based optimisation approaches were developed for dynamic production scheduling, focusing on the Flexible Job Shop Scheduling Problem under realistic disturbances. A reinforcement learning–based scheduling framework was designed and implemented, demonstrating stable training and superior performance compared to heuristic approaches while remaining suitable for real-time use. In parallel, the foundations of the multi-objective optimisation layer were established, formally defining use-case problems, resources, constraints, and conflicting objectives. A modular MO-DDSS architecture was designed to support Pareto-optimal solutions, human-in-the-loop decision-making, and adaptation to disruptive scenarios.
4. The project advanced a comprehensive sustainability assessment of the industrial value chains, covering environmental, economic, and social dimensions. A tailored life-cycle sustainability methodology was defined, supported by detailed value-chain mapping and clear system boundaries. Structured data-collection templates were developed to capture material, energy, cost, and social data at unit level. Stakeholder categories and social impact indicators were identified and validated, resulting in a robust and aligned assessment framework ready for impact analysis and hotspot identification.
5. Human-centred design was strengthened through a systematic analysis of user needs, digital readiness, and training requirements. A Digital Maturity Questionnaire revealed differences between managerial and shop-floor roles and key adoption barriers. Based on these insights, a tiered training structure was defined, and expected learning outcomes were specified. Close collaboration with industrial partners and technology developers ensured alignment between training content, tool complexity, and real operational workflows.
