Risk identification model for power enterprises based on convolutional neural network

This paper proposes a risk assessment model based on Stacking ensemble learning and Convolutional Neural Network (CNN) for power systems with high-penetration renewable energy integration. By constructing wind farm output scenarios for the IEEE 39-bus system, the impact of renewable energy uncertainty on risk identification is systematically analyzed. Experimental results demonstrate that the model achieves optimal performance under 30% renewable energy penetration (accuracy rate: 98.01%, missed detection rate: 2.04%), significantly outperforming single CNN models. The study provides reliable decision-making support for power systems with diverse generation structures.