Research on Deterioration State Evaluation of Hydropower Units based on Successive Variational Mode Decomposition and Mahalanobis Distance

Abstract In order to better acquire the real-time operating status of hydropower units and realize early fault warning, a deterioration state evaluation method for hydropower units based on Successive Variational Mode Decomposition (SVMD) and Mahalanobis Distance (MD) is proposed. In the offline stage, SVMD is optimized with Dispersion Entropy (DE) as the fitness function, and historical health data is used to obtain a health baseline. In the online stage, the real-time monitoring signal is input into the optimized SVMD model first. The features of the Intrinsic Mode Functions (IMFs) are then extracted. Subsequently, Synthetic Detection Index (SDI) and Detection Index (DI) are utilized for feature parameter selection. Finally, a degradation indicator is constructed based on Gaussian mixture model (GMM) and MD, and the degradation curve is drown to evaluate the real-time deterioration state of the unit. Experimental results demonstrate that the proposed method can effectively characterize the real-time operating status of units, identify abnormal changes, and issue timely warnings.