Wavelet LSTM for Fault Forecasting in Electrical Power Grids.

Area of Science:

• Electrical Engineering
• Data Science
• Time Series Analysis

Background:

• Electric power utilities must ensure continuous energy supply.
• Grid failures decrease reliability and performance.
• Accurate fault prediction is crucial for rapid power restoration.

Purpose of the Study:

• To assess the feasibility of time series forecasting for electrical fault prediction.
• To evaluate the long short-term memory (LSTM) model for fault forecasting.
• To investigate the use of wavelet transform to enhance LSTM predictive capabilities.

Main Methods:

• Utilized time series data of electrical power failures in Brazil during 2020.
• Implemented and evaluated the long short-term memory (LSTM) neural network model.
• Integrated wavelet transform with LSTM (wavelet-LSTM) to improve prediction accuracy.

Main Results:

• The wavelet-LSTM model demonstrated superior performance in fault prediction compared to standard LSTM.
• The proposed approach showed reduced prediction error and enhanced robustness.
• Statistical analysis confirmed the model's reliability for practical utility application.

Conclusions:

• Wavelet-enhanced LSTM is a feasible and effective tool for electrical fault prediction.
• This predictive capability aids electric power utilities in optimizing maintenance and improving grid reliability.
• The study validates the utility of advanced forecasting models in critical infrastructure management.