Modélisation qualitative des agro-écosystèmes et aide à leur gestion par utilisation d’outils de model-checking

Yulong Zhao 1
1 DREAM - Diagnosing, Recommending Actions and Modelling
Abstract : The modeling in the domain of agro-ecology is important since it helps us to better understand the interactiosn between the environment and the human activities. Some research works based on simulation has been carried out during the recent years. Mainwhile, not only these simulation tools are difficult to use by the non expert users, but also the high complexity of models makes interactive uses impossible. We propose an approch in which we represent the system to be studied in a discret event system formalism. This kind of representation benefits the efficiency of model-checking and makes it possible to use controller synthesis to generate strategies. We present two contributions in this thesis. The first one concerns the project EcoMata. This project proposes a qualitative modelling which represents a marine prey-predator type food chain in timed automata. Predifined query patterns in natural langurage are also proposed which allow users to investigate easiy the food chain. We have improved the efficiency of the algorithm of timed automata generation and also developped a strategy synthesis method to generate best fishing management strategy. The prototype software EcoMata implements all these propositions including the best strategy synthesis. In the second contribution, we propose a hybrid modelling which represents grazing activities in timed automata. This hybrid modelling combines a numerical grass model and a qualitative grazing model. These sub models are organized in a hierarchical struture of four layers: the biological layer, the activity layer, the decision layer and the clock. We propose four methods to generate best grazing management strategy. One of these methods is applied to the movement of herd. The other three methods are applied to fertilization among which one of them use controller synthesis on timed automata and the other two combine controller synthesis and machine learning to generate generic strategy for a exploitation type. A prototype software PaturMata has been developped which implements this modelling method and the generation of the best strategy of herd movement.
Document type :
Complete list of metadatas

Cited literature [54 references]  Display  Hide  Download
Contributor : Yulong Zhao <>
Submitted on : Monday, February 23, 2015 - 11:55:48 PM
Last modification on : Friday, November 16, 2018 - 1:37:50 AM


  • HAL Id : tel-00933443, version 4


Yulong Zhao. Modélisation qualitative des agro-écosystèmes et aide à leur gestion par utilisation d’outils de model-checking. Ecosystèmes. Université Rennes 1, 2014. Français. ⟨NNT : 2014REN1S008⟩. ⟨tel-00933443v4⟩



Record views


Files downloads