Fuzzy Classification of Internal Combustion Engine Faults Based on Neural Network

Abstract This paper introduces a diagnostic technique for the identification of cylinder power fault degrees in internal combustion engines. After the analysis of the speed fluctuations of an engine flywheel, the faults are divided into several levels corresponding to the characteristic indices of the speed fluctuations by fuzzy set theory. An neural network is used to learn the inconplete fuzzy rules and generalize the implicit fuzzy relations. Performed on-line on an engine test bed, the proposed technique has been proven feasible for identifing the cylinder power fault degrees.