Neuro-fuzzy system for diagnosis of engines, based on oil samples analysis

Abstract

The present paper describes a neuro-fuzzy. hybrid system applied to the diagnosis of automobile engines, based on the analysis of oil samples. A relevance analysls was done to select the most significant variables among the avallable ones, in order to classify the samples. Such relevance analysls is described in detalls along the paper. Four dlfferent systems were implemented one pure neural system, and three dlfferent neuro-fuzzy systems. A detailed descriptlon of the neural and fuzzy systems is also presented, as well as the performance obtained by each one of them.