A ship's system is a complex mechanism composed of multiple pieces of equipment. Due to the dynamic and non-linear characteristics of the parameters of each component, fault diagnosis is complicated. This paper proposes a dynamic feature fusion method for performing efficient fault diagnosis on the system.

Fractal theory, dynamic theory and the kernel principal component analysis (KPCA) method are used to reconstruct, map and filter the system state data, and obtain the principal component characteristic data matrix, square prediction error (SPE) and corresponding control limits. An offline monitoring model based on the health data of a marine diesel engine intake and exhaust system is then constructed and used to diagnose and analyze ship system faults. In order to verify the validity of the model, the fault data of the intake and exhaust system of a marine diesel engine is selected for verification and analysis.

The results show that this method can effectively realize the accurate analysis of a system's dynamic nonlinear state data and efficient analysis and diagnosis of faults, with better fault diagnosis performance than the KPCA and support vector machine (SVM) method.

The method proposed in this paper can realize the detection and diagnosis of marine diesel engine intake and exhaust system failures, and improve the reliability and safety of system operation.