Enhanced forecasting of shipboard electrical power demand using multivariate input and variational mode decomposition...

Area of Science:

• Marine Engineering
• Electrical Engineering
• Data Science

Background:

• Optimizing shipboard Energy Management Systems (EMSs) requires accurate electricity demand forecasting.
• Efficient ship operation and profitability depend on effective EMSs for power grids.

Purpose of the Study:

• To introduce a novel hybrid forecasting approach for shipboard electricity demand.
• To enhance the accuracy of multivariate time series forecasting for energy management.

Main Methods:

• A new Variational Mode Decomposition with Mode Selection (VMDMS) technique was developed for multivariate time series decomposition.
• Long Short-Term Memory (LSTM) networks were employed for forecasting using the decomposed time series.
• The hybrid VMDMS-LSTM model was validated on real-world shipboard power demand data.

Main Results:

• The VMDMS method effectively identifies synergistic modes across channels, improving forecasting accuracy.
• The hybrid approach demonstrated superior performance in predicting shipboard electricity demand.
• The method extends Variational Mode Decomposition (VMD) applicability to multivariate forecasting without restrictive data assumptions.

Conclusions:

• The proposed hybrid forecasting method significantly enhances shipboard electricity demand prediction accuracy.
• This research advances decomposition techniques for predictive modeling in marine energy management.
• The findings offer new pathways for improving the efficiency of shipboard power grids.