Projects / Programmes
Stochastic models for logistics of industrial processes
| Code |
Science |
Field |
Subfield |
| 2.10.02 |
Engineering sciences and technologies |
Manufacturing technologies and systems |
Manufacturing technology |
| Code |
Science |
Field |
| 2.03 |
Engineering and Technology |
Mechanical engineering |
smart factory, optimization of production processes, heuristics and artificial inteligence, production logistics, digital twin
Researchers (20)
Organisations (1)
Abstract
When organizing the production in a modern company, in particular the one that follows the vision of the factory 4.0, the key factor in competitiveness is optimization of production processes. Production planning based on known and well-defined big orders is quickly retiring to the production of small series of products or even to the individualized production, where delivery times are shorter and customers want to be able to change the details of the order as long as possible. Due to this fact and due to the necessity of reduction of storage costs, the stock of materials has to be decreased, which clearly shifts the focus of optimization towards logistics of production systems. The modeling of practical problems with stochastic models has recently been an interesting topic both in theoretical and applied science. The approaches are different, however the gap between theoretical results and practically useful methods is still high. With the proposed research we want to add some original contributions in this segment. This is a research challenge with great potential applicability and is related to the goals of the Slovenian Smart Education Strategy (SVRK, 2015). Within the project, we intend to define and study an original generic model, that will allow to uniformly model the logistics of production processes. We will develop a prototype optimization tool, in which various software modules will be involved that will offer intelligent optimization algorithms for various special models at strategic, tactical and operational levels. Because modules will communicate through the same data structures, natural cooperation of various heuristics will be possible. The aim is to develop heuristic algorithms that will dynamically adapt to changes in the environment, using methods from statistical theory of learning and artificial intelligence. Among the special models we intend to develop, there will primarily be stochastic versions of known and new optimization problems. Solving such tasks is usually computationally intense, so we will use a high-performance computer at FS UL to implement and analyze the heuristics. We are also planning to realize an experimental smart factory in the demo center at FS UL, where we will demonstrate synergic effects due to the participation of optimization modules at different levels. The project is ambitious because expert knowledge and skills from very different fields will be needed, from the very special knowledge on logistics and optimization in smart factory, to deep understanding of the theory of computational complexity and heuristics for hard optimization problems. Knowledge and programming skills in the HPC environment and in the demo center of a smart factory will also be crucial. The project team includes senior researchers and students of mechanical engineering, mathematics and computer science. Their experience and references ensure the feasibility of the project. The foreseen original and innovative results will be, in particular: - Original generic model of production process logistics, which will be general enough to capture many (or all) relevant special models; - Prototype optimization tool with modules for solving optimization tasks at different levels (strategic, tactical and operational); - New models for relevant optimization tasks, where we will mainly focus on stochastic versions; - New heuristic algorithms for new and known optimization tasks, where the novelty will be primarily in the cooperation between heuristics at different levels; - Demonstration of the methodology in the demo center. The original results will be published in scientific literature, and the new knowledge will be incorporated in an appropriate form into the pedagogical process. Prototype software tools and new knowledge will also be useful in future application and development research. The software tool will be designed in a way that will allow later upgrades to enlarge the set of optimization modules. ¦
