2010
Volume 10, Number 1, p.p. 24–32
Toward controlling bioprocesses: a self-adaptive multi-agent system approach
Sylvain Videau,* Carole Bernon and Pierre Glize
Institut de Recherche Informatique de Toulouse, University Toulouse III, 118 Route de Narbonne, 31062 Toulouse cedex 9, France
Controlling complex systems means dealing with their time-varying aspects as well as their nonlinearity. Bioprocesses are an interesting example, due to their complexity and the impossibility of fully describing the microörganisms involved in such reactions, nor their behaviour. In order to overcome these difficulties, we propose an adaptive and generic approach for the control of bioprocesses. This approach relies on the use of an adaptive multi-agent system (AMAS) acting as the controller of a bioprocess. The resulting adaptability enables its application to a wide range of problems and the ability to react to dynamic modifications of the system controlled. The global control problem is thereby turned into a sum of local problems. Interactions between local agents, which solve their own inverse problem and act in a coöperative way, enable the emergence of an adequate global function for solving the global problem while fulfilling the user’s requests. Bioprocesses are often modelled as equation systems, including both differential and classical equations. An instantiation of this approach is applied to an equation-solving problem, and a differential equation-solving problem. The results are detailed and discussed.
Keywords: adaptive control, bioprocess, coöperation, multi-agent systems
PACS: 87.15.R-, 87.15.hp