A data-driven model for power system operating costs based on different types of wind power fluctuations

Area of Science:

• Power Systems Engineering
• Renewable Energy Integration
• Computational Economics

Background:

• Intermittent wind power generation increases power system operating costs.
• Accurately quantifying these costs under complex wind fluctuations is challenging.

Purpose of the Study:

• To develop a power system operating cost model adaptable to diverse wind power fluctuations.
• To precisely map wind energy variability to power system and thermal unit operating costs using a data-driven approach.

Main Methods:

• A two-layer clustering strategy to categorize wind power fluctuations.
• A production simulation model incorporating thermal plant, energy storage, and reserve costs.
• A deep neural network for cost prediction based on wind fluctuation patterns.

Main Results:

• The model accurately simulates overall power system operating costs (4%-18% error).
• It accurately simulates thermal power plant operating costs (3%-13% error).
• The deep neural network effectively maps wind fluctuations to operational costs across seasons.

Conclusions:

• The proposed data-driven model effectively addresses the challenge of calculating operating costs with high wind power penetration.
• The method provides accurate cost estimations, aiding in grid management and economic optimization.
• This approach validates the effectiveness of deep learning in analyzing the economic impact of renewable energy integration.