Detection of knocking in Spark Ignition (SI) engines using CMAC neural networks

Abstract

Knocking in SI (Spark Ignition) engines is one of the most addressable problems. If not detected in early stages, it causes a severe damage to an SI engine. Various techniques have been proposed so far, in order to detect early knock symptoms. This paper presents a novel approach to detect knocking using technique of Artificial Intelligence. A four stroke, single cylinder engine is simulated using GT Power engine simulation software. Data is generated through simulation for both knock and no-knock conditions. A CMAC (Cerebellar Model Articulation Controller) based neural network is then applied as an AI (Artificial Intelligence) tool to distinguish between knock and no-knock conditions. The results show a promising future for CMAC neural networks as a technique to detect knocking in SI engines.