Networks of neurons. Selecting topology and applications for the diagnosis of a DC machine

Abstract

Obtaining a model is a key step for thedecision. All Models residue behaviour in responseto flaws that I have used so far have been establishedthrough a cognitive approach: thus, the useddecision tables resulting either from a knowledgeexpert, (taking into account the signs, deduction ofsequences in simulation), or the study generatorresidue (sensitivity study, qualitative deductionsequences from a transfer). If the use of expertknowledge can be extremely useful, it is not alwaysavailable or effective, and systematic study ofproperties of a generator of waste is often based onrestrictive assumptions. The model of a (dynamic)system may certainly result from a priori knowledge(knowledge model) and of a procedure foridentifying and learning (model behaviour). Ourobjective is thus to build, through learning, adynamic model for the decision.

Keywords

• Networks of neurons
• generator residue
• model dynamic
• generator
• diagnosis
• machine