Renewable Energy Scenario Generation Based on Improved Generative Adversarial Network

With the large number of renewable energy sources such as wind and photovoltaic power integration in the power system, their complex spatial-temporal characteristics bring great challenges to the operation and planning of the power system. The key to overcoming these challenges is to accurately describe the uncertainty of renewable energy sources, which is characterized by its intermittent and random nature. In this paper, the structure of generative adversarial network is improved to simulate the complex spatial-temporal characteristics of wind-solar power. Attention mechanism and temporal convolutional network are introduced to improve the performance for simulating temporal characteristics and spatial correlation. The historical output data of 4 wind farms and 4 photovoltaic power stations in Jiangsu Province in 2022 are used to verify the effectiveness of the proposed method.