A Simulation on Fault Diagnosis Technology with Air and Fuel (A/F) System of Marine Diesel Engine

Abstract

This paper presents a simulation on fault diagnosis technology in signal problems of the air and fuel (A/F) system of marine diesel engine. The research method is used the fault tree analysis (FTA) to analyze the signal problems through expert experiences into a tree diagram and to find out the cause of fault. Then, set the tag to different characteristics, the Kernel Principal Component Analysis (KPCA) is used to reduce the dimensionality and feature extraction of the data, it reduces the computational time and defines the relevance of the fault cause with the alarm sensor. For classification and fault diagnosis technology, the Support Vector Machine (SVM) with optimized characteristics is used to train the model. The experimental results show that the proposed techniques would be improved the accuracy and it will help the marine officers to shorten the debugging time and problem diagnosis time.