Risk identification model for power enterprises based on convolutional neural network.

Area of Science:

• Electrical Engineering
• Artificial Intelligence
• Power Systems Analysis

Background:

• Increasing integration of renewable energy sources introduces uncertainty into power system operations.
• Accurate risk assessment is crucial for maintaining grid stability and reliability with diverse generation structures.

Purpose of the Study:

• To develop and evaluate a robust risk assessment model for power systems with high renewable energy penetration.
• To analyze the impact of renewable energy uncertainty on risk identification using a specific power system model.

Main Methods:

• Proposed a risk assessment model integrating Stacking ensemble learning with Convolutional Neural Networks (CNN).
• Constructed wind farm output scenarios for the IEEE 39-bus system to simulate renewable energy uncertainty.
• Systematically analyzed the model's performance in identifying risks under varying renewable energy penetration levels.

Main Results:

• The Stacking-CNN model achieved an accuracy rate of 98.01% and a missed detection rate of 2.04% at 30% renewable energy penetration.
• Demonstrated superior performance compared to single CNN models in risk identification.
• Validated the model's effectiveness in handling the complexities introduced by renewable energy uncertainty.

Conclusions:

• The proposed Stacking-CNN model offers a reliable approach for risk assessment in power systems with high renewable energy integration.
• The findings provide valuable decision-making support for grid operators managing diverse generation portfolios.
• Highlights the potential of ensemble learning and deep learning techniques in enhancing power system security.