DynamIc control and planning for intelLIGENT manufacturing processes and systems – DILIGENT
DILIGENT: Making Industrial Systems Smarter with AI, Optimization and Digital Twins
Modern manufacturing systems are complex, uncertain and highly interconnected. Traditional approaches treat planning, quality and equipment decisions separately, leading to inefficiencies. DILIGENT addresses this limitation by developing integrated, data-driven and AI-based methods supported by digital twins to improve decision-making and system performance.
Develop integrated, data-driven methods to optimize complex industrial systems and address fragmentation, uncertainty and interdependencies in decision-making.
Modern manufacturing systems, especially in semiconductor production, are highly complex and difficult to manage. They involve many interconnected operations, machines and decisions, all operating under uncertainty. In such environments, traditional approaches are often insufficient because they treat key decisions separately, such as production planning, transport management, quality control and equipment monitoring. One of the main issues addressed by the DILIGENT project is this fragmentation of decision-making. When these decisions are optimized independently, the overall system performance can be degraded. The project aims to overcome this limitation by developing integrated approaches that consider these different aspects simultaneously. Another important issue is uncertainty. Industrial systems are subject to variability in processing times, machine behavior and operating conditions. This makes decision-making more difficult and requires methods that can adapt to changing conditions. DILIGENT addresses this by incorporating data-driven approaches that use industrial data to improve predictions and support more robust decisions. Scalability is also a key challenge. Industrial systems are large and complex, and solving decision problems efficiently requires advanced methods capable of handling large amounts of data and constraints within reasonable computation times. In this context, the main objective of the project is to develop new methods that combine optimization, artificial intelligence and simulation to improve decision-making in complex production systems. The project also aims to better exploit industrial data, improve robustness to uncertainty and provide tools that can be applied in real industrial environments.
The DILIGENT project combines several complementary methods to improve the management of complex industrial systems.
First, optimization techniques are used to better organize production activities and allocate resources efficiently. These methods help identify good solutions even in complex situations with many constraints.
Artificial intelligence is also used to analyze industrial data. Machine learning models allow better prediction of system behavior, such as processing times or machine performance, which helps improve decision-making.
Simulation plays an important role in the project. It allows researchers to reproduce how industrial systems operate and to test different strategies without affecting real production. This helps identify the most effective solutions before implementation.
The project also uses probabilistic models to better take into account uncertainty and variability in industrial environments.
One of the key innovations of DILIGENT is the use of digital twin technologies. These are virtual models of real industrial systems that can be used to monitor, analyze and improve their performance in real time.
By combining these approaches, the project provides new tools to make industrial systems more efficient, more robust and more adaptable to changing conditions.
The DILIGENT project has led to several important results for improving industrial systems.
First, the developed methods have made it possible to better coordinate production and transportation activities. This has led to significant efficiency gains, including strong reductions in transportation times while maintaining good overall performance.
The project also demonstrated the value of using data and artificial intelligence to improve decision-making. By better predicting system behavior, it becomes possible to make more reliable and robust decisions, even in uncertain environments.
Another key result is the integration of different aspects of the system, such as production planning, quality control and equipment monitoring. This integrated approach helps reduce risks and improve overall system performance.
The development of simulation tools and digital twins also represents a major outcome. These tools allow testing different strategies, analyzing their impact and supporting better decisions without disrupting real operations.
Finally, the results were validated using real industrial data, confirming their relevance for practical applications. The project also contributed to scientific dissemination through publications and presentations in international conferences.
The DILIGENT project stands out by its ability to combine several advanced approaches—optimization, artificial intelligence and simulation—into a unified framework to better manage complex industrial systems. A key feature of the project is its integrated vision, where production planning, transportation, quality control and equipment monitoring are considered together rather than separately. This allows for more realistic and efficient decision-making.
Another important strength lies in the use of real industrial data to validate the proposed methods. This ensures that the developed solutions are not only theoretically sound but also applicable in real-world industrial environments.
The project also introduces digital twin technologies, which provide virtual representations of industrial systems. These tools enable continuous monitoring, scenario testing and improved decision support, making it possible to anticipate issues and optimize system performance.
Looking forward, several perspectives emerge. One major direction is to further improve the integration of different decision processes, including maintenance, in order to provide even more comprehensive and adaptive solutions. Another perspective is the development of real-time decision-making tools capable of handling uncertainty and rapidly changing conditions.
The project also opens new opportunities for strengthening the use of artificial intelligence, for example through learning-based approaches that can continuously improve system performance over time.
In addition, the transfer of the developed methods to industrial systems represents an important next step. Their integration into existing platforms will enable companies to benefit directly from the project’s results.
Finally, the approaches developed in DILIGENT can be extended to other sectors facing similar challenges, such as energy systems, logistics and advanced manufacturing.
Berterottière, L.; Dauzère-Pérès, S.; Yugma, C. Flexible job-shop scheduling with transportation resources. European Journal of Operational Research. 2024, 312, 3, 890-909.
Martin, M.; Dauzère-Pérès, S.; Yugma, C. Scheduling on parallel metrology tools for risk reduction in semiconductor manufacturing. ROADEF 2024-25ème congrès annuel de la Société Française de Recherche Opérationnelle et d'Aide à la Décision, Amiens, France. 4-7 Mars 2024.
Lucas Berterottière, Stéphane Dauzère-Pérès, Claude Yugma, Kwei-Long Huang. Minimizing total travel time in the exible job-shop scheduling problem with transportation resources. European Journal of Operational Research, 2025, 312, pp.890-909.
Following the success of IMAGINE, the consortium aims to extend the integrated decisions from the real-time configuration to the predictive strategy. For intelligent manufacturing processes and systems, dynamic dispatching/scheduling and process/equipment control should be managed in a prognostic way. Production plans are dynamically optimized and used as the foundation for the predictive regulations of the process and equipment. As the production and engineering data are accumulated continuously, the predictive control and planning strategies will be empowered by the deep learning techniques that make sense of big data for accurate and precise previsions. Furthermore, the prognostic decisions will be interpreted with reasonable logic and traced back to the controllable factors for not only comprehending the analytic causality but also enhancing the control and planning dynamics. The methods developed in this project will be validated through the cooperation with local partners.
Project coordination
Galliam Claude Yugma (Laboratoire d'Informatique, de Modélisation et d'Optimisation des Systèmes)
The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.
Partnership
LIMOS Laboratoire d'Informatique, de Modélisation et d'Optimisation des Systèmes
National Taiwan University / Graduade Institute of Industrial Engineering
Help of the ANR 213,996 euros
Beginning and duration of the scientific project:
- 36 Months
