Abstract: This paper is concerning the application of a deterministic decisional pattern to a multi-agent system which would provide intelligence to a distributed energy smart grid at local consumer level. Development of multi-agent application involves agent specifications, analysis, design and realization. It can be implemented by following several decisional patterns. The purpose of present article is to suggest a new approach to control the smart grid system in a decentralized competitive approach. The proposed algorithmic solution results from a deterministic dichotomous approach based on environment observation. It uses an iterative process to solve automatic learning problems. Through memory of collected past tries, the algorithm monotonically converges to very steep system operation point in attraction basin resulting from weak system nonlinearity. In this sense, system is given by (local) constitutive elementary rules the intelligence of its global existence so that it can self-organize toward optimal operating sequence.
Abstract: Energy Efficiency Management is the heart of a
worldwide problem. The capability of a multi-agent system as a
technology to manage the micro-grid operation has already been
proved. This paper deals with the implementation of a decisional
pattern applied to a multi-agent system which provides intelligence to
a distributed local energy network considered at local consumer level.
Development of multi-agent application involves agent
specifications, analysis, design, and realization. Furthermore, it can
be implemented by following several decisional patterns. The
purpose of present article is to suggest a new approach for a
decisional pattern involving a multi-agent system to control a
distributed local energy network in a decentralized competitive
system. The proposed solution is the result of a dichotomous
approach based on environment observation. It uses an iterative
process to solve automatic learning problems and converges
monotonically very fast to system attracting operation point.