Fault detection in the marine engine using a support vector data description method

Abstract

Fast detection and correct diagnosis of any engine condition changes are essential elements of safety and environmental protection. Many diagnostic algorithms significantly improve the detection of malfunctions. Studies on diagnostic methods are rarely reported and even less implemented in the marine engine industry. To fill this gap, this paper presents the Support Vector Data Description (SVDD) method as applied to the fault detection of the fuel delivery system of a two-stroke marine engine. The selected diagnostic data is the exhaust gas composition, with four components considered: oxygen, carbon oxide, nitric oxide, and carbon dioxide. With these diagnostics, the method distinguishes eight different engine faults from the efficient state. The manuscript presents in detail the methodology for applying the SVDD method in a marine engine. The method of obtaining diagnostic data and its scaling is described. The method of training and validating the algorithm is also presented, along with ready-made algorithms for use. The 100% accuracy of the proposed fault detection method. Based on the obtained results, the proposed fault detection method is promising for a simple application. Moreover, generalised algorithms that may be adapted to different technical solutions are also presented. Highlights SVDD was used for marine engine fault detection from exhaust gas composition Laboratory measurements were carried out on the two-stroke diesel engine The proposed algorithm detected the considered faults with 100% accuracy A generalised algorithm for adapting other complex technical objects was proposed