Team 3: Smart Computing and management of complex systems

Smart Computing et pilotage des systèmes complexes .

Team involved
Name of the project manager :
Lotfi-Nabli - Professor
Name of teacher-researchers involved Grade
Hédi dhouibi Assistant Master
Ouni Khaled Assistant Master

Name of doctoral students to be mobilized within the framework of the project:

Marwa GAM

Amir ABDELLATIF

Nesrine BLEL

Marwa DORAI

Ibtissem BEN GHARAT

Asma BASLY

Hichem KMIMECH

Salah HAMMEDI

The research program is situated in the context of complex and discrete event systems which will be mainly modeled by Petri nets (Petri nets). The work will concern, on the one hand, the development of tools allowing the extraction of knowledge from databases (Data Mining) of complex systems (transport systems, food industry, medical systems, industrial systems, etc. ). We develop hybrid approaches exploiting statistical tools (PCA and nonlinear PCA; Bayesian approach); signal processing tools (Wavelet transform and wavelet networks) and intelligent tools (Fuzzy logic, Neural networks, genetic algorithms, ant colony, belief function, multi-agent approaches, etc.). On the other hand, the issue of resource allocation is one where the continuity of production is a common objective, the aim of which is to define a production orientation procedure. An approach based on Petri Networks (MOP) modeling the problem of resource allocation of multi-cycle systems is introduced and a development of the conceptual and theoretical framework. A proposal for heuristics and meta-heuristics in order to reduce the complexity of theoretical computation will be developed.

The objective is the establishment of cooperative systems in transport in general, and the development of new traffic control strategies in urban networks. We will have to study, develop and simulate an access control strategy at intersections in an urban traffic network, using communicating cooperative systems. The implementation of an urban traffic simulation software package where it is planned to set up tools for the modeling of v2v (vehicles to vehicles) and v2i (vehicles to infrastructure) communications. In addition, a contribution to the development of the process for determining intelligent tools to ensure the operational safety of transport systems by setting up monitoring / control and risk management strategies ensuring safe operation of road circuits (safety systems). transport).


Amélioration du système de production

Digital tools in current production systems allow monitoring and optimization of the overall performance of installations. The solutions proposed must contribute to the reduction of the cost of ownership (unavailability, cost of maintenance). In addition, in order to protect or contribute to the collaborative protection of a set of systems (eg: group of robots or robots / equipment), generic multi-model architecture solutions for diagnostics must be proposed.


Need for health monitoring:

Improving customer services by stepping up assistance to operators and the use of degraded modes is only possible through more effective monitoring of the various functions, in particular “critical” functions. The management of the information collected by a function monitoring system (HUMS: Health and Usage Monitoring System) allows live prognosis and reconfigurations. Thus, the anticipated management of failures will provide continuity of service to users. This monitoring of system parameters could also promote the anticipation of degradations to avoid latent failure..


Maintenance Need :

For maintenance, the operation of the monitoring system through the communication between the maintenance PC and the state of health of the production system, must allow flexible use of the wear indicators of the elements concerned. The anticipation of maintenance tasks is ensured thanks to the implementation of an interactive software for monitoring and helping to find faults. In addition, scheduling maintenance tasks and sourcing spare parts helps reduce equipment downtime. The best localization of errors thanks to an improved diagnosis will then make it possible to avoid successive removals and therefore to reduce maintenance costs.

Optimizing a set of complex resource allocation problems while relying on labeled Petri Nets. Thus, for some complex systems, the Ptri Network structure (RdP) is giant, the application of the approaches proposed in literature seems impossible in terms of computation time which increases exponentially. In fact, we are firstly contributing to the design of an approach (meta heuristic which allows the optimization of the resource allocation problems of multicycles systems based on RdP, such as monitoring-command problems at base of mobile robots This approach consists primarily of optimizing the availability of resources, hence the control of resources under constraints implicitly linked to other parameters such as cost, processing times and pre-allocations.

The research program is entitled:Smart Computing et pilotage des systèmes complexes.

Research work is particularly oriented towards the design and development of intelligent tools to aid the safe and optimal driving of complex systems. They will provide contributions to the development of processes for determining intelligent tools, knowledge extraction, statistics or signal processing to ensure the dependability of complex systems by:

  • The design and development of monitoring strategies, direct and indirect, with or without a model for the monitoring - detection and localization functions for the diagnosis, prognosis and implementation of a maintenance strategy.
  • The design and development of strategies for the control (modeling and simulation, Robustness) and / or monitoring of the different types of processes taking into account the performance indicators of each.
  • The integration of control strategies through intelligent approaches to monitoring (Monitoring of control law parameters)
  • Risk analysis and management ensuring safe operation of distributed systems using multi-agent type modeling approaches.
  • The development of protection against all of these classes of threats by detecting potentially critical situations in order to avoid catastrophic failures.
  • Security techniques and methods for complex systems in an uncertain and imprecise environment, by using main component analysis tools and / or artificial intelligence tools,
  • Taking diversification into account by distinguishing two processing chains: the first supports functional processing and the second is intended for the control of the functional chain.
  • Our task is to show that the chain of custody verifies a set of rules deduced from the security needs of the system. Thus, overall operational reliability depends on the effectiveness of these rules, and on the representativeness of the underlying assumptions in relation to real situations.
  • We show that unlike classical automatic systems, autonomous systems perform multiple tasks dynamically selected according to the goals to be achieved. The security rules to be verified can therefore change depending on the tasks performed.

As part of the research work of the team, we propose to define the different sets of rules within “intelligent security modes”. In addition, we will define a formal notation for the specification of these modes.

Tokyo

Tokyo is the capital of Japan.