Extraction of Situational Data for Electricity-Gas Integrated Energy System Using Deep Learning Methods

This study adopts a deep learning-based approach to extract situational data for an integrated electricity-gas energy system, aiming to enhance performance and operational efficiency by offering real-time situational awareness. The process starts by using state estimation techniques to determine the state and deviation vectors of the energy system at the situational perception level. These vectors reflect the real-time status of the system, highlighting discrepancies between predicted and actual system behavior. At the situational comprehension level, a recurrence plot (RP) method is applied to analyze historical deviations in system performance. RP serves as a powerful tool for visualizing time-series data, helping to identify complex patterns in the system's deviations over time. Finally, at the situational projection level, deep learning techniques like Convolutional Neural Networks (CNN) are employed to forecast future operational states based on historical deviation data, represented by recurrence plots. The CNN model is trained to predict the system's future behavior, thus improving the accuracy of forecasts and risk management in the integrated energy system. This approach offers a strong framework for extracting and projecting situational data, facilitating better decision-making and ensuring more efficient system operation.

کلیدواژه ها:

data-driven approach ، electricity-gas integrated energy system ، convolutional neural network