A Novel Fault Diagnosis Algorithm for the Fuel Injection System of Marine Two-Stroke Diesel Engine

Abstract

The fuel injection system plays an important role in guaranteeing the diesel engine’s power and emission along with its economical and safe usage. Current research focuses mainly on the four-stroke diesel engine, where fault diagnosis is conducted through measuring the running parameters of the fuel injection system followed by the analysis of the running conditions via feature extraction. Few studies focus on fault diagnosis for the fuel injection systems of two-stroke diesel engine. Thus, this paper investigates this domain by using three artificial intelligence algorithms: the back-propagation network, the self-organizing feature map network and support vector machine. Experimental results on the marine two-stroke diesel engine demonstrate that the support vector machine can classify the working conditions more accurately and efficiently in the case of limited samples. The accuracy rate of fault diagnosis can reach 95% when the parameters of this algorithm are optimized via the grid optimization method.