Modèle multi-agents d'aide à la décision pour la gestion des services préhospitaliers d'urgence

Abstract

The need for a better understanding and good control of the growing complexity of information systems requires the development of new methods of modeling and problemsolving. In this research study, we are interested in the design and modeling of a decision support system in which the knowledge and skills of the expert enables the analysis and proposing of new multi-agent models. The development of such a model identifies a number of design challenges, linked in particular to the efficiency and effectiveness of the calculation process of the solution. Many complex systems are characterized by non-linear dynamics, both disordered and random; in summary, complicated in the sense that their assimilation requires time and talent. Traditional mathematical methods (differential equations, probability models, etc.) may be inappropriate for the modeling of such systems, in which interaction plays a very important role. Modeling based reactive agents is one of the most prevalent microscopic modeling techniques. Why choose an agent-oriented modeling rather than another modeling metamodel? First, the agent model is very rich. It helps to easily schematize qualitative and quantitative processes and allows for interacting heterogeneous entities in various architectures. However, the main reason is often related to the modeling vocation: to fully understand the relationship between an individual and a collective action/behavior. This work is undertaken primarily in an application context related to the problem of the planning and management of prehospital emergency services (PES). Indeed, there is a series of research efforts that address the topic of the management and planning of PES. Each research work addresses a very specific problem in the field, such as scheduling paramedics, managing demand, or the management of vehicles/ambulances, etc. This thesis focuses on the planning problem of PES to reduce the response time of paramedics. It adopts a global and integrated resolution approach. It proposes a model in the form of various components of decision support. It incorporates optimization technologies that involve paramedics scheduling, managing staff replacements, managing vehicle fleets, the management of deposits capacity, coverage of demand, and management of special events. The proposed model is based on a multi-agent architecture and can meet the constraints and unforeseen events encountered in the planning of PES. This thesis is structured around the following three papers: • "Integrated and global approach (IGAP) based on multi-agent systems for the management of prehospital emergency services", submitted to Computers & Industrial Engineering of Elsevier. This paper presents an overview of multi-agent systems applied to prehospital emergency services and proposes a new global and integrated approach for its resolution called IGAP. • "Scheduling System for Prehospital Emergency Services", submitted to the European Journal of Operational Research of Elsevier. This article addresses the problem of paramedics scheduling. Our contribution is the proposal of a mathematical model that solves a coverage problem integrated in a new system of scheduling that is sufficiently flexible. • "Multi-Agent Decision-Making Support Model for the Management of Prehospital Emergency Services", published in the International Journal of Machine Learning and Computing of IACSIT. This article focuses on the topic of modeling and decision support in the context of complex systems and proposes architecture based agents for the planning and management of prehospital emergency services.

Date	25 Jun 2015
Original language	French
Awarding Institution	École de technologie supérieure
Supervisor	Marc Paquet (Supervisor)