Marielle Delhom

Water management policy selection using a decision support system based on a multi-agent system

In this paper, we present the framework of a decision support system for water management in the Mediterranean islands coupling a multi-agents system with a geographic information system. The platform developed makes it possible for users to better understand the current operation of the system, to apprehend the evolution of the situation, and to simulate different scenarios according to the selected water policies and the climatic changes hypothesis. We present the model and the simulations we ran to select the best rules to control the consumers during a period of water shortage.

Prediction of the hydrologic behavior of a watershed using artificial neural networks and geographic information systems

The authors have been working on the modeling and simulation of environmental systems. We have developed an object oriented approach mainly based on discrete events and object oriented simulation. In order to validate our approach, we applied it to the simulation of watershed behavior; the results were quite satisfactory. However, a more accurate rainfall/runoff model is necessary in order to take into account hourly flow variations, particularly for flood prediction. As our first approach was not efficient enough to do this, we chose to define a new one based on artificial neural networks (ANNs) and geographic information systems (GIS) paradigms.

Analyzing knowledge exchanges in hybrid MAS GIS decision support systems, toward a new DSS architecture

In this paper we present the strong points and the weakness of GIS used as decision support systems. We expose the advantages of adding MAS to GIS. We present the common critters used to characterize the MAS-GIS links, and introduce a new point of view about these relations studying the nature of the knowledge used to help decision makers. We explicit the nature and the purpose of these knowledge, and we study the knowledge exchanges between MAS, GIS, experts and decision makers. Then, we propose a new architecture, SMAG, for hybrid MAS-GIS decision support systems development platforms. We implement this architecture and build a decision support system dedicated to the fresh water problem. Thus we illustrate the relevance of our operational decision making solution that is of one the first to make possible to take into account together individual and social behaviors, spatial component and specificities of target sites.

Water Management Using a New Hybrid Multi-Agents System - Geographic Information System Decision Support System Framework

In this paper we first present the problems at the origin of our works to propose a generic decision support system framework based on multi-agents system (MAS) and geographic information system (GIS). We explicit why we chose to use a multi-agents approach to make models and what advantages brings the use of a geographic information system. Then we present our SMAG generic architecture for MAS-GIS based decision support system (DSS), and CORMGIS the first SMAG compliant platform. Finally we present the fresh water problematic in the Mediterranean island and the experimentations we made using CORMGIS to study the impact of social influence and spatial agents location on the dynamic of the system during dry periods.

Definition of a decision support system to control a sheep herd

The ovine breeding is still a vital activity in a number of areas all over the world. Under the pressure exerted by the European Community the farmers are constrained with heavy investments. These additional loads strongly burden the profitability of the traditional exploitations. They impose an evolution in the control of the herds which implies the use of new food, the quality control of milk and the meat, and genetic amelioration of the animals. Faced with this extremely significant cost, we propose to develop a decision support systems based on the modeling of the farms and the simulation of the various types of control of the herds. This kind of system allows: to reduce the number of the experiments and their duration, to highlight the key factors of a farm control of and to define practices having best profitability.

Hierarchical Multi-Views Modeling Concepts for Discrete Event Systems Simulation

In this article, we describe a Hierarchical Multi-Views Modeling Concepts for Discrete Event Systems Simulation. We give the basic concepts needed for the definition of the formal model allowing discrete events simulation of complex systems [1].

A Smart System to Support Proximity Based Solidarity in Corsica SmartAngels

— The purpose of this paper, is to provide a solution to the issue of dependence for elderly people in the context of an ageing population phenomenon. In rural areas with a low population density, isolation has negative impacts on everyday life. In France, the ageing population raises the fear of a serious increase in the number of persons suffering from disabilities and in need of assistance. In Corsica, population ageing is substantially increasing. Given this critical context we are trying to develop a system of connected objects called SmartAngels to support proximity-based solidarity. It consists of a meshed network of autonomous ZigBee microcontrollers, in which various actors can exchange information by means of an accurate interface design. Its main task is to support both proximity-based solidarity and social relationships in rural environments. Moreover, the SmartAngels device grid ensures optimal neighbourhood solidarity in case of a serious problem by creating a community of actors. First aid can be provided by neighbouring people alerted by the system, thus allowing to efficiently handle the distance to civil protection assistance, and to overcome technical limits in areas where telecommunication networks are either underdeveloped or saturated. In this work, we present a SmartAngels system prototype allowing to handle the social link and launch alerts in rural environments with a low density of population.

Hybrid MAS GIS Mediterranean backcountry tourism economy modeling methodology

Submitted to the combination of the sun immigration pressure and the decline of their traditional economy, the Mediterranean islands are in a paradoxical position. Indeed, while the populations on the shores are growing, the countryside is endangered involving the desertification. In this paper we shall present a methodology to build a distributed artificial intelligence based model dedicated to the countryside economy and its interactions with the growing touristic coastal areas through the flow of tourists. Organizing the economic system into a hierarchy, following a multi-agent approach, we build a generic model of each systems stakeholders from our observations. We propose a methodology to model the exchanges between the countrysides visitors and the local economys players. Using a geographic information system, we explicit how to take advantage of the spatiotemporal ground data to fit the simulations to the specificities of each targeted site. Finally we present a first application on a Corsican case.

An hybrid methodology for the modeling and the simulation of natural systems

In this paper, we first present concepts we developed in order to define an hybrid modeling and simulation methodology dedicated to the study of natural systems.These concepts are based on the object-oriented design, artificial neural networks and the geographic information systems. Then, we outline a general framework combining these three concepts.A first application allowing the forecasting of the hydrologic behavior of watersheds is described and some results are presented and commented.