电网技术2018，Vol.42Issue(6)：1860-1867,8.DOI:10.13335/j.1000-3673.pst.2018.0291
风电和光伏随机场景生成的条件变分自动编码器方法
Conditional Variational Automatic Encoder Method for Stochastic Scenario Generation of Wind Power and Photovoltaic System
摘要
Abstract
Along with increasing penetration of renewable energy sources, such as wind power and photovoltaics, their volatility has a adverse influence on stable and economic operation of power grids, so it is more important to model uncertainty of renewable energy. Random scenario analysis is a major method to solve the problem. Existing methods build probability model and generate scenarios based on sampled historical data. But major obstacles remain, such as low model accuracy and high computational complexity. This paper proposes a stochastic scenario generation method based on conditional variational autoencoder (VAE). Compared with existing methods, the proposed method can carry out unsupervised learning for temporal, spatial, and fluctuant characteristics of training data, and generate scenarios efficiently matching characteristics of observation data according to the given condition without scenario reduction. Effectiveness of the proposed algorithm is verified for three testing scenarios: single unit, multiple units and designated tag scene.关键词
随机场景分析/条件变分自动编码器/深度学习/场景生成Key words
stochastic scenario analysis/conditional variational automatic encoder/deep learning/stochastic scenario generation分类
信息技术与安全科学引用本文复制引用
王守相,陈海文,李小平,舒欣..风电和光伏随机场景生成的条件变分自动编码器方法[J].电网技术,2018,42(6):1860-1867,8.基金项目
国家电网公司科技项目(52153217000F).Project Supported by Science and Technology Project of State Grid Corporation of China (52153217000F). （52153217000F）
