Adaptive Supervision of Patterns in Discrete Event Systems: Application to Crisis Management

Abstract

Crisis management is currently an important challenge for medical service and research. This motivates the development of new decision system approaches to assist (or to guide) the decision makers. A crisis management is a special type of collaboration involving several actors. The context and characteristics of crisis such as extent of actors and their roles make the crisis management more difficult in order to take decision. In this paper, we propose to model the interaction between different actors involved in crisis management. For this purpose we use finite state automaton in order to optimize the emergency response to the crisis and to reduce the disastrous consequences on people and environment. Thus, an adaptive supervision method is proposed. Therefore, we address the problem of diagnosis and prediction (prognostic) given an incomplete model of the discrete event systems of a crisis situation. When the model is incomplete, we introduce learning into the diagnoser (diagnosis module) construction.