Imbalance Fault Classification Based on VMD Denoising and S-LDA for Variable-Speed Marine Current Turbine

Jiajia Wei, Tao Xie, Ming Shi, Qianqian He, Tianzhen Wang, Yassine AmiratJournal of Marine Science and Engineering 2021 · DOI

TL;DR

This paper introduces a novel method for classifying blade imbalance faults in marine current turbines (MCTs) by combining Variational Mode Decomposition (VMD) for denoising and Screening Linear Discriminant Analysis (S-LDA) for classification. This approach effectively identifies fault severity even under challenging, variable marine conditions, achieving a high accuracy of 92.04%.

What Problem Does This Paper Address?

Marine current turbines (MCTs) are crucial for harnessing renewable energy from ocean currents. However, they are susceptible to various faults, with blade imbalance being a significant concern. This type of fault can reduce power generation efficiency and potentially damage the turbine system. A major challenge in diagnosing these faults is the harsh marine environment, characterized by unstable currents and background noise, which makes it difficult to accurately classify the severity of blade imbalance. Existing methods often struggle with non-stationary conditions and the complexity of real-world data, necessitating a robust and reliable fault classification technique.

Methods

The proposed method integrates three key techniques:

  • Hilbert Transform (HT): This is used for phase demodulation of the single-phase stator current signal. HT helps extract the instantaneous frequency, which is a crucial indicator of faults.
  • Variational Mode Decomposition (VMD) Denoising: VMD is employed to decompose the signal into several intrinsic mode functions (IMFs). This decomposition allows for effective denoising by isolating the IMF that contains the fault characteristic frequency (1P frequency) and removing noise and other interfering components. This is a key step in handling the "seafloor noise" and current instability mentioned.
  • Screening Linear Discriminant Analysis (S-LDA): After denoising and analyzing the power spectral density (PSD) of the relevant IMF, S-LDA is used for fault classification. S-LDA is an enhanced version of traditional LDA that includes a pre-screening step. This pre-screening, combined with the power signal analysis, aims to improve classification accuracy and efficiency by focusing on the most informative features and reducing the influence of irrelevant data.

Datasets

The study utilized experimental data collected from a 230-W marine current turbine prototype. Different severities of blade imbalance faults were simulated by attaching varying masses (0g for healthy, 8% and 15% imbalance) to the blades. The experiment was conducted under variable marine current velocities (1.4 to 1.6 m/s), simulating non-stationary conditions. A total of 3000 sample points were collected for each data set, with 380 healthy samples and 380 fault samples for each class.

Key Insights and Ideas

  • Stator Current as a Reliable Indicator: The research highlights that even when stator currents appear similar in the time domain across different fault severities, subtle changes in their frequency characteristics, particularly when analyzed after denoising, can reveal critical fault information.
  • VMD's Power in Noise Reduction: VMD proves to be highly effective in isolating the fault signature from the complex and noisy marine environment, a significant challenge for traditional methods.
  • S-LDA's Advantage in Feature Selection: The "screening" aspect of S-LDA, by pre-selecting relevant features (IMFs with fault characteristics), significantly enhances classification accuracy compared to standard LDA, especially in complex, real-world scenarios.
  • Robustness to Velocity Fluctuations: A crucial finding is that the proposed method effectively eliminates the influence of variable flow velocity on fault classification, a common issue in marine environments.

Key Results

The proposed method achieved a fault classification accuracy of 92.04% for distinguishing between healthy, 8% imbalance, and 15% imbalance faults. This accuracy was maintained even under variable marine current velocities. The study also compared its performance against traditional LDA, K-Nearest Neighbor (KNN), and Support Vector Machine (SVM), demonstrating superior accuracy over LDA (33.33%) and competitive results against KNN (87.22%) and SVM (87.96%).

Limitations and Future Work

  • Experimental Scale: The experiments were conducted on a relatively small-scale prototype (230-W). The applicability of the method to larger, industrial-scale MCTs would require further validation.
  • Other Fault Types: The study focuses specifically on blade imbalance faults. Future work could explore extending this methodology to detect and classify other types of MCT faults, such as bearing or gearbox issues.
  • Real-World Data Complexity: While the experimental setup simulated non-stationary conditions, real-world marine environments can present even more complex and unpredictable challenges. Further testing with diverse field data would be beneficial.
  • Parameter Optimization: The paper mentions the use of VMD and S-LDA, but detailed optimization strategies for their parameters in the context of MCT fault diagnosis could be a direction for future research to further enhance performance.

Summary updated

Abstract

Marine current energy as a kind of renewable energy has gradually attracted more and more attention from many countries. However, the blade imbalance fault of marine current turbines (MCTs) will have an effect on the power production efficiency and cause damage to the MCT system. It is hard to classify the severity of an MCT blade imbalance fault under the condition of the current instability and seafloor noise. This paper proposes a fault classification method based on the combination of variational mode decomposition denoising (VMD denoising) and screening linear discriminant analysis (S-LDA). The proposed method consists of three parts. Firstly, phase demodulation of the collected stator current signal is performed by the Hilbert transform (HT) method. Then, the obtained demodulation signal is denoised by variational mode decomposition denoising (VMD denoising), and the denoised signal is analyzed by power spectral density (PSD). Finally, S-LDA is employed on the power signal to determine the severities of fault classification. The effectiveness of the proposed method is verified by experimental results under different severities of blade imbalance fault. The stator current signatures of experiments with different severities of blade imbalance fault are used to validate the effectiveness of the proposed method. The fault classification accuracy is 92.04% based on the proposed method. Moreover, the experimental results verify that the influence of velocity fluctuation on fault classification can be eliminated.
BibTeX
@article{wei2021imbalance,
  author = {Jiajia Wei and Tao Xie and Ming Shi and Qianqian He and Tianzhen Wang and Yassine Amirat},
  title = {{Imbalance Fault Classification Based on VMD Denoising and S-LDA for Variable-Speed Marine Current Turbine}},
  journal = {Journal of Marine Science and Engineering},
  year = {2021},
  doi = {10.3390/jmse9030248},
}
Figure