A multi-mode process monitoring method based on mode-correlation PCA for Marine Current Turbine

Under random occurrence of turbulence and wave, the power generation process of Marine Current Turbine (MCT) exists multiple operating modes. The multi-mode characteristics and frequent change between modes make the fault detection difficult. To monitor multi-mode processes, a mode-correlation Principal Component Analysis (PCA) method is proposed. Firstly, the multi-mode characteristics of MCT are presented. Then, the paper gives the effect of multi-mode process on the conventional PCA and analyses the relationship between the modes. Finally, a normalized method is proposed to dynamically fit the mode, the statistic value between different modes in PCA is removed. Theoretical analysis and experimental results verify the effectiveness of the proposed method.