A Convolutional Neural Network Based on Channel Cross Attention for Efficient Probabilistic energy Flow Calculation in the Integrated Electrical and Natural Gas System

The increasing proportion of renewable energy installed capacity in integrated energy systems(IES) has led to a corresponding rise in uncertainties. Probabilistic Energy Flow (PEF) analysis, while effective in evaluating these uncertainties, is often hindered by high computational costs. This paper proposes a fast PEF calculation method using Convolutional Neural Networks (CNNs) to address this challenge. The study introduces two key contributions: (1) a feature construction that captures renewable energy fluctuations, multi-load uncertainties, and network topology changes in IES, and (2) a Channel Cross Attention Module (CCAM) for adaptive feature selection and fusion in power and gas systems. Finally, the effectiveness of the proposed method is verified on the modified IEEE 14-NGS14 system.