The title of the postdoc's research topic is as follows: “Development of metaheuristics to assist in the proactive parameterization of the DDMRP method.”
The origin of the topic stems from an Idex Transformation 2020 project (Excellence Initiative). The Transformation project consists of a one-year experiment to transform existing course sets and associated teaching practices in order to improve the success and professionalization of students. The amount granted is €19,976. This project is supported by the Master of SCM of EMSBS and INSA Strasbourg.
The Idex Transformation project
The teaching practice to be experimented with is the active teaching method (see Celestin Freinet). This teaching practice should enable students to challenge the models described in the scope of the project. Questioning the models involves the description of concepts, the interaction between concepts, the formalization of concepts into management models, and finally the evaluation of the management models. The challenged models must correspond to the models generally used in companies. Evaluating models is delicate without a simulation phase (serious game or software tools). Learning is self-regulated. Experimental trial and error, or scenario development, is carried out by exploring problem situations, information and interaction situations, and solution situations. A certain amount of feedback is delivered via the learning environment itself. The teacher becomes a resource in the learning environment.
In order to concretely launch the experiment in the current year, a 317-page textbook has been written: Supply Chain Management, des Origines A DDMRP / Une analyse conceptuelle et détaillée des méthodes de planification et de pilotage à l’aide d’un ERP.
The postdoc project
Drafting the textbook within the framework of the Idex Transformation project has made it possible to identify gaps in the scientific literature concerning the parameterization of the DDMRP method. The difficulty lies in the fact that the parameterization of planning and steering methods is contextual. This means that there are no general parameterization rules, only trends. The parameterization solutions are first empirically determined, then refined using the trial and error method. These approaches are limited and do not guarantee optimal parameterization.
The postdoctoral research project consists in developing a method to assist parameterization using a computerized optimization process. The process makes good use of metaheuristics and, in particular, of recent hybrid metaheuristics approaches. Preliminary results are encouraging. In order to validate the robustness of the developed method, experiments on industrial data are underway.