Research on Online Diagnosis of Ship Main Engine with Unbalanced Sample

Abstract

The traditional ship main engine fault diagnosis model is difficult to update in real time. In addition, the main engine also has the problem of more monitoring parameters and less fault samples. In order to eliminate the influence of the above problems, improve the diagnostic accuracy. Firstly, in order to solve the problem that the traditional diagnosis model is difficult to update online in real time, Online Sequence Extreme Learning Machine is adopted as the diagnosis model, which can update the model directly online by using real-time samples. Then, for the problem that the sample set collected in the online update process is much smaller than the normal sample set, BorderlineSMOTE is used to generate fault samples to balance the sample set, and feature extraction and dimension reduction are carried out in combination with KPCA to further improve the training speed and diagnostic accuracy of the model. Taking the main engine fuel system as an example, the experimental results show that the proposed algorithm has high diagnostic accuracy and good stability, and is suitable for the research of ship main engine fault diagnosis.